/V

.:t:;1QA Prfj~~rtj of r:~ n~~~r~r:!fr:~ ITt

TH~f/iE': :~,:~L ;\'.;~~ :\f-'~' :_~~i;' ;::t~>~;':j\:t~C3

#~B5
CIVIL ENGINEERING STUDIES
STRUCTURAL RESEARCH SERIES NO. 285

STUDIES F THE FATIGUE BEHAVI R

F BU ElDED J INTS
M6tZ
IN HY-80 AND HY-100 STEELS
Reference Ro~m
Civil Engineering Department
BI06 C. E. Building
University of. Illinois
Urbana, Illinois 61
By
W. H. MUNSE
W. H. BRUCKNER
A. J. HARTMANN
J. B. RADZIMINSKI
R. W. HINTON
and
J. L MOONEY

A REPORT OF AN INVESTIGATION CONDUCTED

by
THE CIVIL ENGINEERING DEPARTMENT
UNIVERSITY OF ILLINOIS
in cooperation with
THE BUREAU OF SHIPS, U.S. NAVY
Contract NObs 88058
Project Serial No. SR-007-01-01, Task 856

UNIVERSITY OF ILLINOIS
URBANA, ILLINOIS
NOVEMBER, 1964
 Property of t!;9 reDsmme~t of
THEOHEriC:H i::;;; .~t;:;:_iLC ;'~iLGiiAriIGS
UNi~'I~~Sn ~ Df !LLifl~OiS

STUDIES OF THE
FATIGUE BEHAVIOR OF BUTT-WELDED JOINTS
IN HY-80 AND HY-100 STEELS

by
W. H. Munse
W. H. Bruckne r
A. J. Ha r't rna nn
J. B. Radzirninski
R. W. Hinton
and
J. L. Mooney

A Report of an Investigation Conducted

by

THE CIVIL ENGINEERING DEPARTMENT

UNIVERSITY OF ILLINOIS

In Cooperation With
The Bureau of Ships, U. S. Navy
Contract NObs 88058
Project Serial No. SR-007-01-01, Task 856

UNIVERSITY OF ILLINOIS

URBANA, I LL INO I-S

NOVEMBER 1964
 SYNOPSI S

The results of axial fatigue tests of transverse butt welded joints

in HY-80 steel containing internal defects are reported. The tests were con-

ducted on 3/4 in. and 1-1/2 in. thick material under stress cycles of zero-to-

tension and complete reversal. The types of weld defects studied include

porosity, slag inclusions, and artificially produced pores, and show that the

existence of weld defects measurably decreases the fatigue 1 ife of a joint at

a given stress cycle when compared to sound welded specimens (with reinforcement

removed). However, there does not appear to be a simple correlation between

type of defect or amount of defective area and fatigue strength.

As part of the study of weld defects, radiographic and ultrasonic

techniques were employed to determine the time to initiation and propagation

of fatigue cracks in welded joints. The results indicate that the first

appearance of a macrocrack (visible on the radiographs) occurs at sl ightly

more than half the test 1 ife. The subsequent crack propagation progresses

slowly and erratically at first, but the rate of growth is rapidly accelerated

from the time it reaches the specimen surface until complete fai lure occurs.

Metallurgical studies were conducted to examine the effect of restraint

during welding on the formation of internal microcracks (sulfide inclusions) in

welded joints, and the behavior of these joints under conditions of fatigue.

The studies indicate that the effects of metallurgical factors introduced by

restraint during welding are less critical than internal weld flaws (porosity,

slag, etc.) in reducing fatigue strength.

A preliminary study of the fatigue behavior of 3/4 in. thick HY-100

steel was completed. Tests were carried out on plain plate specimens and trans-

verse butt welded joints in the as-welded condition using a zero-to-tension stress
cycle. Although the results for the plain plate specimens compared favorably

with similar tests on HY-80 steel, the fatigue behavior of the welded joints

in HY-100 was inferior to that of the HY-80 material. This is apparently due

to a greater sensitivity of the HY-100 steel to external weld geometry and to

the presence of even very small weld defects.

 -iv-

TABLE OF CONTENTS

INTRODUCTION.

1.1 Object of Study. 1

1.2 Scope of Investigation 1
1.3 Acknowledgments . . . . . . 2

i I. DESCRIPTiON OF TEST PROGRAM 3

2.1 Material 3
2.2 Fabrication of Specimens 3
2.3 Description of Test Equipment. 4
2.4 Testing Procedure . . . 4
2.5 Radiographic Studies 4

I I I. FATIGUE TESTS OF 3/4 iNCH HY-100 MATERIAL 6

3.1 I ntroductory Remarks 6

3.2 Tests of Plain Plate Specimens 6
3.3 Tests of As-Welded Butt Joints 7
3.4 Data Analysis . . . . . . . 8

IV. FATIGUE TESTS OF TRANSVERSE BUTT WELDS IN HY-80 MATERIAL

HAVING INTERNAL DEFECTS 9

4.1 Introductory Remarks 9

4.2 Porosity in 1-1/2 In. HY-80 Specimens. 9
4.3 Slag in 1-1/2 In. HY-80 Specimens. 12
4.4 Fatigue Tests of Defective Welds in 3/4 In.
HY - 80M ate ria 1 . . . . .... 13
4.5 Discussion of Results . . . 14

V. CRACK INITIATiON AND PROPAGATION. 15

5.1 Introductory Remarks ..... 15

5.2 Test Equipment and Testing Procedure 15
5.3 Initiation of Fatigue Cracks 16
5.4 Crack Propagation . . 19
5.5 Conclusions . . . . . 22

VI. METALLURGICAL STUDIES OF WELDED HY-80 STEEL 24

6.1 Introductory Remarks . . . . . . . 24

6.2 Experimental Procedures 25
6.3 Material Description 26
6.4 Distribution and Size of Sulfide Inclusions 26
6.5 Discussion . . . . . . . . . . . . . . . . 27
a. Hot Crack Geometry . . . . . . . . . . . 28
b. Effect of Restraint of the Butt Weld on Hot Cracks 28
c. Effect of Sulfur Concentration on Hot Cracking 29
 -v-

TABLE OF CONTENTS (cont'd)

d. Length of Hot Cracks . . . . . .... 29

e. Metallographic Observations of Weld Metal Flaws 29
f. Inclusion Study . . . . . . . . . . 30
g. Effect of ilButter i ngll on the Frequency of Hot Cracks 30
h. Effect of Hot Cracks on Fatigue Life . . . . . 31
6.6 Conclusions. 33

VI I. SUMMARY 35

7.1 Fatigue Behavior of HY-100 Steel 35

7.2 Weld Defect Study of HY-80 Steel 35
7.3 Crack Growth Study . . . . . . 36
7.4 Metallurgical Studies . . . . . . 36
7.5 General Summary and Conclusions. 37

BIBLIOGRAPHY 39

TABLES . 40

FIGURES

APPENDIX A: Ultrasonic Crack Detection Study A-l

 -vi-

LIST OF TABLES

Number

2.1 PHYSICAL PROPERTIES OF BASE METAL. 40

2.2 CHEMICAL COMPOSITION OF BASE METAL 41
3.1 RESULTS OF FATIGUE TESTS OF HY-100 PLAIN PLATE SPECIMENS 42
3.2 RESULTS OF FATIGUE TESTS OF HY-100 TRANSVERSE BUTT WELDS
IN THE AS-WELDED CONDITION . . . . . . . . . . . . . 43
3.3 SUMMARY OF FATIGUE TESTS OF HY-80 AND HY-100 MATERIAL 44
4.1 RESULTS OF FATIGUE TESTS OF TRANSVERSE BUTT WELDS IN
1-1/2 IN. HY-80 STEEL (Intentional Porosity in Weld) 45
4.2 RESULTS OF DEFECT EXAMINATION OF TRANSVERSE BUTT WELDS IN
1-1/2 IN. HY-80 STEEL (Intentional Porosity in Weld) 46
4.3 RESULTS OF FATIGUE TESTS OF TRANSVERSE BUTT WELDS IN
1-1/2 IN. HY-80 STEEL (Intentional Slag Inclusions in Weld) 47
4.4 RESULTS OF DEFECT EXAMINATION OF TRANSVERSE BUTT WELDS IN
1-1/2 iN. HY-80 STEEL (Intentional Slag Inclusions in Weld) 48
4.5 RESULTS OF FATIGUE TESTS AND DEFECT EXAMINATION OF TRANSVERSE
BUTT WELDS IN 3/4 IN. HY-80 STEEL (!ntentional Porosity and
Slag) . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.1 RESULTS OF FATIGUE CRACK INITIATION AND PROPAGATION STUDY
OF TRANSVERSE BUTT WELDS IN 3/4 iN. HY-80 STEEL CONTAINING
INTENTIONAL WELD DEFECTS . . . . . . . . . . . . . . . 50
6.1 MANGANESE, SULFUR, AND PHOSPHORUS CONTENT OF THE HY-80
STEEL PLATES . . . . . . . . . . . . . . . . . . . . . 51
6.2 AVERAGE SIZE OF SULFIDE INCLUSIONS PER NUMBER MEASURED 52
6.3 NUMBER OF HOT CRACKS PER TRANSVERSE SECTION OF HY-80
STEEL EXAMINED . . . . . . . 52
6.4 AVERAGE LENGTHS OF HOT CRACKS . . . 53
6.5 LENGTH OF HOT CRACKS NOT ASSOCIATED WiTH INCLUSIONS. 53
A.l RESULTS OF ULTRASONIC STUDY OF CRACK PROPAGATION A-2
 -y i i-

LIST OF FIGURES

Numbe r

2.1 DETA I LS OF TEST SPEC IMENS

2.2 WELDING PROCEDURE P80-11018-A
2.3 WELDING PROCEDURE P100-11018-J
2.4 WELDING PROCEDURE P80-11018-D
2.5 ILLINOIS l FATIGUE TESTING MACHINE AS USED FOR AXIAL LOAD ING OF
WELDED JOINTS
3.1 TYPICAL PHOTOMICROGRAPHS OF TRANSVERSE BUTT WELD IN 3/4 IN. HY-80
MATERIAL
3.2 TYPICAL PHOTOMICROGRAPHS OF TRANSVE RSE BUTT WELD IN 3/4 IN. HY-I00
MATER IAL
3.3 RESULTS OF FATiGUE TESTS OF AS-ROLLED HY-100 PLAIN PLATE SPECIMENS
3.4 FRACTURE SURFACE OF AS-ROLLED HY-I00 PLAIN PLATE SPECIMEN
3.5 RESUL.TS OF FATiGUE TESTS OF HY-100 TRANSVERSE BUTT WELDS IN THE
AS-WELDED CONDITION
3.6 TYPICAL FRACTURES OF TRANSVERSE BUTT WELDS IN 3/4 IN. HY-100 MATERIAL
3.7 REDUCTION IN FATIGUE STRENGTH FROM AS-ROLLED PLAIN PLATES DUE TO A
TRANSVERSE BUTT WELD IN THE AS-WELDED CONDITION
4.1 RESULTS OF FATIGUE TESTS OF 1-1/2 IN. TRANSVERSE BUTT WELDS CONTAINING
VARIOUS AMOUNTS OF INTENTIONAL POROSITY (REIN~ORCEMENT REMOVED)
4.2 TYPICAL FRACTURES FOR 1-1/2 IN'. TRANSVERSE BUTT WELDED JOINTS
CONTAINING INTENTIONAL INTERNAL WELD DEFECTS
4.3 PERCENT FATIGUE STRENGTH AT 100~000 CYCLES VS. EFFECTIVE DEFECT
DIAMETER OF POROS ITY IN 1-1/2 IN':'. MATER IAL
4.4 COMPUTED FAT I GUE STRENGTH AT 100,000 CYCLES VS. DEFECT IVE AREA DUE TO
INTENTIONAL POROSITY IN 1-1/2 IN. MATERIAL
4.5 RESULTS OF FATIGUE TESTS OF 1-1/2 IN. TRANSVERSE BUTT WELDS CONTAINING
VARIOUS AMOUNTS OF INTENTIONAL SLAG (REINFORCEMENT REMOVED)
4.6 COMPUTED FATIGUE STRENGTH AT 100,000 CYCLES VS. DEFECTIVE AREA DUE TO
iNTENTiONAL SLAG iNCLUSiONS iN 1=1/2 IN. MATERIAL
4.7 COMPUTED FATIGUE STRESGTH AT 20~000 CYCLES VS. DEFECTIVE AREA DUE TO
INTENTIONAL SLAG INCLUSIONS IN 1-1/2 IN~ MATERIAL
4.8 FRACTURE SURFACE OF 3/4 IN. TRANSVERSE BUTT WELDED JOINT CONTAINING
ARTIFICIAL INTERNAL WELD DEFECTS
4.9 COMPUTED FATIGUE STRENGTH AT 50,000 CYCLES VS. DEFECTIVE AREA DUE TO
INTENT 10NALWELD DEFECTS IN 3/4 IN. MATERIAL
4.10 COMPUTED FATIGUE STRENGTH AT 100,000 CYCLES VS. DEFECTIVE AREA DUE
TO 'INTENTIONAL WELD DEFECTS IN 1-1/2 IN'. MATERIAL
5.1 SETUP FOR RAD,I O'GRAPH IC STUDY OF CRACK PROPAGAT ION
 -v iii-

LIST OF FIGURES (CONTD)

Numbe r

5.2 SETUP FOR ULTRASONIC STUDY OF CRACK PROPAGATION

5.3 CORRELATION BETWEEN INITIAL LENGTH OF INTENTIONAL WELD DEFECT AND
INITIATION OF FATIGUE CRACKS
5.4 COMPARISON OF CYCLES OF FATIGUE CRACK PROPAGATION TO TEST LIFE IN
3/4 IN. HY-SO TRANSVERSE BUTT WELDED SPECIMENS TESTED AT 50.0 KSI
5.5 COMPARISON OF CYCLES OF FATIGUE CRACK PROPAGATION TO TEST LIFE IN
3/4 IN. HY-SO TRANSVERSE BUTT WELDED SPECIMENS TESTED AT 0 TO
+50.0 KSI
5.6 RATE OF CRACK PROPAGATION IN 3/4 IN. HY-SO TRANSVERSE BUTT WELDED
SPECIMENS TESTED AT 50.0 KSI
5.7 RATE OF CRACK PROPAGATION IN 3/4 IN. HY-SO TRANSVERSE BUTT WELDED
SPECIMENS TESTED AT 0 TO +50.0 KSI
6.1 WELDING JIG USED TO RESTRAIN HY-SO STEEL DURING BUTT WELDING
6.2 MAIN PLATE OF HY-SO STEEL FROM WHICH TEST PLATES G-67, G-6S 9 AND
G-69 WERE OBTAINED
6.3 SECTIONING PLAN FOR METALLOGRAPHIC SAMPLES OF BUTT WELDED HY-SO STEEL
6.4 TRANSVERSE SECTION OF A MULTIPASS BUTT WELD OF HY-SO STEEL
6.5 HISTOGRAM OF THE WIDTH OF INCLUSIONS IN THE Z-S7 and G-69 SAMPLES OF
HY-SO STEEL
6.6 HISTOGRAM OF THE THICKNESS OF INCLUSIONS IN THE Z-S7 AND G-69 SAMPLES
OF HY-SO STEEL
6.7 PHOTOMICROGRAPH OF A LARGE CRACK IN THE HAZ AND WELD METAL OF BUTT
WELDED HY-SO STEEL
6.S PHOTOMICROGRAPH OF THE SAME CRACK SHOWN ABOVE AFTER 0.009 IN-. WAS
POLISHED FROM THE TRANSVERSE SECTION
6.9 PHOTOMICROGRAPH OF A LARGE CRACK NOT APPARENTLY ASSOCIATED WITH
INCLUSIONS
6.10 PHOTOMICROGRAPH OF CRACKS ASSOCIATED WITH POROSITY NEAR UNDERCUT
RE GION 0 F WE LD
6.11 PHOTOMICROGRAPH OF A LARGE CRACK ALONG THE "FINGER" OF THE WELD METAL
THAT EXTENDS INTO THE BASE METAL
6.12 MICROHARDNESS SURVEY OF 1-1/2 IN. THICK HY-SO STEEL BUTT WELD
6.13 MI CROHARDNESS SURVEY OF THE WELD METAL AND HAZ OF "BUTTERED " HY-SO
STEEL SCARF
6.14 FRACTURED SE'CTION OF HY-SO STEEL SAMPLE G-6S WELDED IN RESTRAINT
6.15 TRANSVERSE SECTION OF FRACTURED SAMPLE INDICATED BY THE ARROW IN
FIG. 6.14
6.16 PHOTOMICROGRAPH SHOWING FATIGUE CRACKS RADIATING FROM POROSITY IN
HY-SO STEEL SAMPLE G-6S AFTER 13,000 CYCLES OF 50.0 KSI
 -ix-

LIST OF FIGURES (CONTD)

Number
A.l RESULTS OF ULTRASONIC CRACK DETECTION STUDY FOR SPECIMENS TESTED
AT 50.0 KSI
A.2 RESULTS OF ULTRASONIC CRACK DETEG.TION STUDY FOR SPECIMENS TESTED
AT 0 TO +50.0 KSI
 I. INTRODUCT ION

1.1 Object of Study

(1 2) -;',
Since the previous studies' on fatigue behavior of welded joints

in HY-80 steel sometimes produced failures that initiated in the weld at.

internal flaws rather than at changes in the external geometry, an extensive

study of the fatigue behavior of 3/4 in. thick defective welded joints in

HY-80 steel was undertaken in July 1962. In the present study, the effect of

internal weld flaws on the fatigue behavior of 1-1/2 in. thick HY-80 welded

joints has been evaluated and compared to that obtained for 3/4 in. specimens.

In addition, the study of initiation and propagation of fatigue cracks was

expanded to study the effect of defect type and stress cycle.

The second phase of the study has been concerned with a prel iminary

evaluation of the fatigue behavior of HY-IOO steel. Two types of specimens were

studied: as-rol led plain plate specimens, and butt welded joints in the

as-welded condition.

The third phase of the study involved a metallurgical study of the

effect of internal cracks (which initiate at sulfide inclusions), produced by

restraint during welding, on the fatigue behavior of welded joints.

1.2 Scope of Investigation

The studies reported herein were carried out on plain plate and

transverse butt welded specimens using stress cycles of zero-to-tension and

complete reversal. The report covers tests conducted during the period of

July 1963 to July 1964 and presents the resul ts of a number of large scale

fatigue tests.

Numbers in parentheses refer to corresponding entries in bibl iography.

 -2-

Sixteen tests of 1-1/2 in. butt welded joints having slag and/or

porosity were carried out under zero-to-tension stress cycles. In addition, a

number of tests of 3/4 in. butt welded joints were conducted to study the

effect of stress cycle (zero-to-tension and complete reversal), and type of

internal weld flaw on the fatigue behavior of welded joints. These specimens

were tested with the weld reinforcement removed.

Fatigue tests were carried out on plain plate and butt welded speci-

mens fabricated from 3/4 in. HY-100 material. A zero-to-tension stress cycle

was used for these tests.

Fatigue tests of butt welds fabricated under high welding restraint

were used to study the effect of restraint on the development of internal weld

cracks and their subsequent effect on the fatigue behavior of butt welded

joints in HY-80 material.

1.3 Acknowledgments

The tests described in this report are the result of an investigation

conducted in the Engineering Experiment Station of the University of III inois.

The program was carried out with funds provided by the Bureau of Ships, U.S.

Navy, under Contract 88058, Project Serial No. SR-007-01-01, Task 856.

The investigation constitutes a part of the structural research

program of the Civil Engineering Department of which Dr. N. M. Newmark is the

head. The research was conducted by A. J. Hartmann, J. B. Radziminski, and

J. L. Mooney, Research Assistants in Civil Engineering, under the direct super-

vision of W. H. Munse, Professor of Civil Engineering. The metallurgical

studies were conducted by R. W. Hinton, Research Assistant in Metallurgical

Engineering, under the direct supervision of W. H. Bruckner, Research Professor

of Metallurgical Engineering.

The authors wish to express their appreciation to all the people on

the Staff of the Universi,ty who so ably assisted in the investigation.

 -3-

I!. DESCRIPTION OF TEST PROGRAM

2.1 Ma te ria 1

Two steels of different strength, HY-80 and HY-100, were used in

this study; the HY-80 material was used in 3/4 in. and 1-1/2 in. thicknesses

while the HY-100 material was used in 3/4 in. thickness. The physical and

chemical properties of the materials are given in Tables 2.1 and 2.2, respec-

tive1y.

The welding electrodes employed throughout the test program were of

Mil 11018 grade. The electrodes were conditioned according to the requirements

of the Navy Specifications(5); a complete description of the procedure is

give n i n Re f. ( 2) .

2.2 Fabrication of Specimens

Figure 2.1 shows the geometry of the specimens tested in this test

program. The specimen blanks, 9 by 48 in., were flame cut from larger sections

of as-ro1 led HY-80 and HY-100 steel plates. The blanks used to fabricate

transverse butt welded joints were cut in half and the sawed edges were

beveled at 30 degrees to provide a double-V groove with a 60 degree included

angle. The welding procedures that were used are given in Figs. 2.2 and 2.3

(3/4 in. thick material) and Fig. 2.4 (1-1/2 in. thick material). All

welding was done in the flat position while the specimen blanks were clamped

to a special welding jig. A stringer bead technique was used in placing

successive passes and provides a heat treatment to the previously placed

passes.

The holes necessary to hold the specimen in the testing machine

were drilled and then the specimen blanks were machined to the desired
 -4-

configuration (Fig. 2.1). In no case was material near the test section

removed by flame cutting. After being machined, the edges of the specimens

were ground and pol ished.

2.3 Description of Test Equipment

The tests were carried out using the University of III inois '

250,000 lb. lever-type fatigue machines (Fig. 2.5); one of the machines was

operated at 100 cpm while the other was operated at 160 cpm. There is a 15 to

force mul tipl ication provided between the dynamometer and the test specimen by

the lever system.

The throw of the eccentric determines the load range; i.e., the

difference between the maximum and minimum loads. The value of the maximum

load is set by means of the turnbuckle which is situated just below the

dynamome te r.

2.4 Testing Procedure

After the load had been set and the machine started, a microswitch

was set so that the machine would shut off when a crack had propagated ~rtially

through the specimen. Failure was assumed to have occurred when the micro-

switch shut the machine off. The microswitch was then disconnected and the

machine restarted; this was done to completely fracture the specimen so that

the fracture surface could be examined and photographed.

2.5 Radiographic Studies

All specimens were subjected to radiographic examination prior to

test i ng in order to determine the qual i ty of the transverse butt wel ds. All

"sound" weld specimens showed no weld defects to be present although a number

of the specimens failed in the weld at minute weld flaws.

 -5-

In addition, radiographs were taken at various stages of the tests

of some 3/4 in. butt welded specimens of (HY-80) purposely provided with slag

and porosity to evaluate crack initiation and propagation.

 -6-

! I I. FATIGUE TESTS OF 3/4 INCH HY-100 MATERIAL

3.1 Introductory Remarks

Fatigue tests were carried out on plain plate and as-welded butt

joint specimens of HY-100 material. During the past few years, the use of

HY-100 has increased and as a result a knowledge of its fatigue behavior has

become important.

The transverse butt welds were fabricated using welding procedure

P100-1 l018-J, which is identical to procedure P80-1 1018-A used for the HY-80

material; no difficulty was encountered in the welding of the HY-100 material.

The weldor and the welding procedure were qual ified in accordance with

requirements given in Ref. (5) for HY-80 material. The joint was able to
0
undergo a 180 guided side bend test without cracking. Figures 3.1 and 3.2

show typical micrographs for the HY-80 and HY-100 materials, respectively;

the microstructures of these materials are very similar.

3.2 Tests of Plain Plate Specimens

The results of the fatigue tests of plain plate specimens on a

zero-to-tension stress cycle are tabulated in Table 3.1 and plotted in Fig. 3.3.

The S-N curve shown in this figure for HY-100 material was obtained using the

procedure out1 ined in Ref. (2). In addition, Fig. 3.3 contains the S-N curve

for 3/4 in. HY-80 plain p1ates(4). Comparing the curves it is noted that the

slope for the HY-100 material is greater than that for the HY-80 material.

However, only a 1 imited number of tests were included in the study.

In a number of tests, failure initiated at the mill scale surface

near the radius of the test section rather than at the radius of the test

section. These failures initiated about 1/8 to 1/4 in. from the machined edge;
 -7-

Fig. 3.4 shows this type of failure. The results tabulated in Table 3.1

indicate that the test 1 ives were not influenced by the mode of failure. One

of the specimens, HY-5, failed due to fretting in the upper pullhead after

396,500 cycles. The fretting cracks initiated at the drilled holes in the

pul lheads; examination of the area around the holes showed that the edges of

the holes were very rough. A new pullhead was welded to the specimen and a

total 1 ife of 546,300 cycles was obtained before failure occurred.

Comparing the average S-N curves for the HY-lOO and HY-80 material,

one may note that although the fatigue strength of HY-lOO is greater than the

fatigue strength of HY-80 for lives of less than 150,000 cycles, the situation

is reversed at longer lives.

3.3 Tests of As-Welded Butt Joints

The resul ts of the fatigue tests of as-welded butt joints on a

zero-to-tension stress cycle are tabulated in Table 3.2 and plotted in Fig. 3.5.

Figure 3.5 also contains the S-N curve for 3/4 in. HY-80 as-welded butt
.. (4)
J 0 I nts . A1 though the 3/4 in. HY-80 spec i mens all failed due to the stress

concen t ra t ion of the ex te rna 1 wel d geomet ry, this was not true for the HY-lOO

material. In a number of cases, failure occurred in the weld rather than at

the edge of the weld reinforcement because of very smal I defects which were

present in the root area of the weld. These defects (Fig. 3.6) were not visible

on the radiographs taken of the as-welded specimens prior to testing. Even on

the fracture surface, it was difficult to locate the defects because of their

smallness.

It can be seen on Fig. 3.5 that there was a large amount of scatter

in the test results when failure occurred in the weld. In addi,tinn the failures

propagated rapidly when they originated in the weld. When failure occurred in
 -8-

the weld, the fatigue 1 ives were less than those for HY-80 material by a

factor of about 8.

Even when failure was due to the external weld geometry, the fatigue

behavior of the HY-100 material was inferior to that of the HY-80 material.

Between 40,000 and 400,000 cycles the 1 ives of the two materials, at a given

stress level, varied by a factor of 2 to 3. The difference increased with

increas ing 1 ife.

3.4 Data Analysis

Table 3.3 1 ists the fatigue strengths and slopes of the S-N curves

for 1-1/2 and 3/4 in. HY-80 and 3/4 in. HY-100 materials.

Al though the fatigue strength of the HY-100 plain plate specimens is

sl ightly greater than that of the HY-80 for short 1 ives (less than 100,000

cycles), it is sl ightly less than that of the HY-80 for longer 1 ives. In the

case of transverse butt welds, the behavior of the HY-100 joints is inferior

to that of the HY-80 joints and the slopes of the S-N curves for HY-100 material

are larger than the corresponding slopes of the HY-80 material, indicating

that the HY-100 steel is more notch sensitive in fatigue than HY-80.

Figure 3.7 shows the reduction in fatigue strength resulting from

the inc 1 us ion 0 f the we 1de d j 0 i n t s a s a fun c t ion 0 f 1 i fe for the HY- 8 and
HY-100 steels. The decrease in fatigue strength due to welding HY-100 is

35 percent at N = 50,000 cycles and 43 percent at N = 200,000 cycles. For

3/4 in. HY-80 material, the reduction was 8 percent at N = 50,000 cycles and

19 percent at N = 200,000 cycles, again showing the increased fatigue notch

sensitivity of the HY-100 steel.

The HY-100 butt welded joints were very sensitive to the presence of

internal defects. No tests were conducted with the weld reinforcement removed

since most welds would contain minute weld defec~s and 1 ittle or no increa.se in

1 ife could be expected.

 -9-

IV. FATIGUE TESTS OF TRANSVERSE BUTT WELDS IN HY-80 MATERIAL

HAVING INTERNAL DEFECTS

4.1 I ntroductory Remarks

Fatigue tests were carried out on a number of defective butt welded

specimens in two thicknesses of HY-80 steel (3/4 in. and 1-1/2 in.) using two

stress cycles (complete reversal and zero-to-tension). The weld reinforcement

was removed on all specimens to improve defect definition on the radiographs

taken of each specimen and to remove any effect of the weld reinforcement on

the fatigue behavior.

The porosity and slag defects in the 3/4 in. thick specimens were

produced as described in Ref. (4). However~ some experimentation was necessary

to produce the larger amounts of slag and porosity desired in the 1-1/2 in.

specimens. This was necessary because the increased thickness required larger

defect areas to provide defects of the maximum size permitted by Navy

Specifications(5) .

In addition, several 3/4 in. thick transverse butt welded specimens

were fabricated with artificial defects (steel balls) in the root area of the

weld. The procedure had been reported previously(4).

4.2 Porosity in 1-1/2 In. HY-80 Specimens

The results of fatigue tests of 1-1/2 in. defective butt welds are

tabulated in Table 4.1 and plotted in Fig. 4.1. Since the tests were carried

out using two stress cycles, 0 to +70.0 ksi and 0 to +50.0 ksi, it was possible

to draw the S-N curve which is shown in Fig. 4.1. The slope, k = 0.21, is
larger than that for sound butt welds with weld reinforcement removed but is

less than that for butt welds tested in the as-welded condition. The S-N

curves for sound butt welds with and without weld reinforcement are also given
 -10-

in Fig. 4.1 Comparing the three curves one may see that although the curve

for defective welds 1 ies entirely below that for welds with reinforcement

removed, it is below the curve for as-welded joints for 1 ives less than

150,000 cycles but above the curve for higher I ives. This indicates that at

long 1 ives the external geometry is more critical, with respect to fatigue,

than internal defects when the defects are of a size permitted by the code;

at the shorter 1 ives the internal defects produce greater effects than the

external geometry.

The geometry of the porosity in the 1-1/2 in. thick material was

similar to that of the 3/4 in. thick material which was discussed extensively

in a previous report(4). The majority of natural defects were roughly spherical

or ellipsoidal ~ Fig. 4.2a. The minimum distance from the surface of the weld

to the porosity ranged from 7/16 in. to 9/16 in. for the 1-1/2 in. thick

material and 7/32 in. to 5/16 in. for the 3/4 in. thick plate. All specimens

in the present test series with natural porosity were fabricated to produce a

single cluster of porosity. Specimens G-8l and G-88 each had fatigue cracks

that grew from the main cluster of porosity as weI I as from other small defects

located some distance from the cluster. Specimen G-8l had the lowest 1 ife

and specimen G-88 had the longest 1 ife of the specimens tested at 0 to +50.0 ksi.

The pore diameters in most of the test welds ranged from 1/64 in. to
-'-
3/64 in. The effective defect diameter", d , was determined for the largest
e
flaw in any cluster that propagated to failure and is tabulated in Table 4.2.

Figure 4.3 shows a plot of the percent of F 000 for a sound weld with the
IOO ~

reinforcement removed vs. the effective defect diameter; over the small range

Effective defect diameter de =.fd d , where d and d = dimensions of major

l 2 1 2
and minor axes of pore. The ~aximum value of d is 1 isted in Tabl~ 4.2
e
for all specimens.
 -11 -

of values studied there does not appear to be any consistent variation in

fatigue strength with the defect parameter, d This is in contrast to the

e
results obtained for 3/4 in. material where an increase in effective defect

diameter produced a decrease in F 000' Nevertheless, it may be seen that

100 ,
the welds with maximum defects ranging from 0.035 to 0.067 in. in diameter

had fatigue strengths at 100,000 cycles of approximately 65 percent of that

for a sound weld.

Another parameter that has been used to specify the amount of porosity

is the percentage of defective area in the gross area of the specimen, A d '
re .
The defective area is defined as the area of porosity in the group or groups

propagating to failure as measured on the fracture surface or as measured from

the radiograph assuming that the porosity is uniformly distributed in a

s phe rica 1 manne r . The A d for a specimen was usually somewhat greater when
re .
based on the fracture surface than when based on radiographs because of the

difficul ty of interpreting the radiographs.

The average A d for the 1-1/2 in. thick specimens was approximately
re .
equal to 0.16 percent; this is approximately the same as that obtained for

the 3/4 in. thick specimens tested previously. The parameter, A

re d. ' (based
on the radiographs) is plotted vs. F 000 in Fig. 4.4. There is a 1 ittle
100 ,
scatter in the test results, and no consistent decrease in fatigue strength

with increasing A . is evident for 0.02 < A . < 0.25 percent. This is
red red
unl ike the case for 3/4 in. material for which a trend of decreasing fatigue

strength with increasing defective area was evident.

Table 4.2 also indicates whether welds satisfy Naval Specifications(5).

Only two of the members failed to meet the Specification requirements but all

of the members exhibited fatigue strengths considerably below that of a

sound weld.
 -12-

4.3 Slag in 1-1/2 In. HY-80 Specimens

The results of fatigue tests of transverse butt welds having slag

inclusions in the weld are tabulated in Table 4.3. Since the tests were

carried out using two stress cycles, it was again possible to obtain an S-N

curve for the test results. The test results are shown in Fig. 4.5. In

addition to the S-N curve for the defective welds, Fig. 4.5 contains the S-N

curves for sound transverse butt welds in the as-welded condition and with

the weld reinforcement removed. It is notable that the slope of the S-N curve

for welds with slag inclusions is the same as that for porosity specimens

(k = 0.21). (One of the specimens, G-90, failed due to the presence of a

small slag deposit although the radiograph indicated only porosity.)

Figure 4.6 shows the fatigue strength at 100,000 cycles as a function

of the percentage of defective area, A d (based on radiographs). The values

re .
of A d ' based on radiographs and on fracture surfaces are tabulated in
re .
Table 4.4; this table also indicates whether the welded joints satisfy Navy

Specifications(5). Unl ike the case for 3/4 in. thick material where an

increase in defect area was accompanied by a decrease in fatigue strength,

there does not appear to be a similar trend for 1--1/2 in. material. As shown

in Fig. 4.6 the zero-to-tension fatigue strength at 100,000 cycles varies

between +37 ksi and +50 ksi, independent of the value of A d ' while the
re .
average fatigue strength for as-welded transverse butt welded joints is +46 ksi

and for sound welded joints with the reinforcement removed is +64 ksi. The

scatter for as-welded joints is +38 ksi to +48 ksi, which is essentially the

same as fo r we 1ds with in te rna 1 defec ts . I n some cases, we 1ded j 0 in ts not

meeting the weld qual ity requirements of the Navy Specifications had higher

fatigue strengths than those which did meet the requirements (see specimens

G-91 vs. G-92 and G-93 in Table 4.3, and specimens Z-96 and Z-94 vs. Z-56, Z-74

and Z-95 in Table 4.5).

 -13-

When the fatigue strength at 20,000 cycles is considered, Fig. 4.7,

the effect of internal geometry on the fatigue 1 ife becomes more important.

At this 1 ife, the upper bound of the results for defective welds is about 6 ksi

lower than the average fatigue strength of the sound as-welded joints.

4.4 Fatigue Tests of Defective Welds in 3/4 .J.n. HY-80 Material

Three transverse butt welded specimens, Z-lOO, Z-lOl, Z-102, containing

artificial porosity (drilled holes filled with steel balls), were tested to

study the effect of the spacing and number of defects in aline on the fatigue

1 ife. However, it was not possible to obtain fully the desired information

since Z-lOO failed due to lack of fusion in the root area rather than at

drilled holes. The 1 ives for these three specimens are given in Table 5.1 of

the propagation study.

Specimen Z-102, which had eleven 1/16 in. diameter bal Is spaced at

approximately 3 hole diameters, had a relatively long fatigue 1 ife for a

specimen with A d equal to 1.2 percent. This can be attributed to the widely
re .
distributed defect area and to the shape of the defects. Only six of the steel

balls were included in the initial fatigue portion of the crack which propagated

to failure (Fig. 4.8).

A number of 3/4 in. butt welded joints having internal weld defects

were tested using a 50.0 ksi stress cycle. The test results are tabulated in

Table 4.5; some of the complete reversal tests previously reported in Ref. (4)

are also included. The fatigue strength at 50,000 cycles has been plotted as

a function of the percentage of defective area, A d ' in Fig. 4.9. It does

re .
not appear that the fatigue strength is a function of the defect area

Furthermore, the upper bound of the scatter band for defective welds is about

12 ksi below the average fatigue strength for sound as-welded butt joints.
 -14-

4.5 Discussion of Results

The fatigue strengths at 100,000 cycles vs A d for 1-1/2 in.

re .
transverse butt welds shown in Figs. 4.4 and 4.6 have been summarized in

Fig. 4.10. The graph indicates that, for the range of intentional defects

studied, the effect on the fatigue strength is independent of the type of

defect and the amount of defective area. It is evident also that at 100,000

cycles the fatigue strength of the members with intentional flaws is about

the same as that of the as-welded joints with the reinforcement in place (no

intentional defects).

From the study reported above, it appears that the intentional internal

weld defects in specimens with the weld reinforcement removed are of more

importance than the external geometry of sound as-welded joints at shorter 1 ives,

less than 100,000 cycles, while the external weld geometry appears to be more

important at the longer 1 ives. It is also evident that the present code

requirements for porosity and slag will not guarantee a fatigue behavior for

1-1/2 in. thick butt joints comparable to that of a sound joint. Neither does

there appear to be any correlation, within the specification$, between fatigue

strength and the defect size or area visible on radiographs.

Unl ike the resul ts obtained for zero-to-tens ion tests of defective

3/4 in. butt welded joints, there does not appear to be any change in the

fatigue strength at a given 1 ife with an increase in theperceDtage of defective

area, A
. re'd . ' for complete reversal tests .
 -15-

V. CRACK INITIATION AND PROPAGATION

5.1 Introductory Remarks

The fatigue 1 ife of metals may be divided into two general categories:
_'~

the number of cycles required to initiate a fatigue crack" or cracks at a

given stress level, and the cycles of crack propagation from initiation to

failure. As part of the present study, nine 3/4 in. thick HY-80 transverse

butt welded specimens, Z-94 through Z-102 inclusive, were examined during

testing by using both radiographic and ul trasonic techniques to determine the

time to crack initiation and subsequent propagation. Five specimens were

tested at 50.0 ksi while the remaining four were tested at 0 to +50.0 ksi.

Each of the specimens contained weld defects, the defects varying in type and

size as shown in Table 5.1. The data for three specimens from an earl ier

study(4) , Z-70, Z-79, and Z-80, are also included in the results.

5.2 Test Equipment and Testinq Procedure

All specimens in this series were tested in the University of

III inois ' 250,000 lb. fatigue machine described earl ier. To study crack

initiation and propagation, two separate techniques were followed. The first

employed the use of the portable radiographic equipment shown in Figs. 5.1a

and 5. lb. The second method, used primarily for a qual itative comparison with

the radiographic technique, util ized the ul trasonic flaw detection equipment

shown in Figs. 5.2a and 5.2b.

Each of the specimens was x-rayed prior to testing to determine the

size and location of internal weld defects. The specimen was next secured in

Crack initiation - estimated he-rein as one-hal-f- the number of cycJes

between last radiographic observation showing no crack and first radio-
graph showing crack.
 -16-

the fatigue machine and subjected to repeated loadings (the number of cycles

depending on stress cycle) at which time the machine was shut off with the

specimen being held at maximum tensile load. With the specimen in this

position, a radiograph was then taken of the test section. Fol lowing this,
0
the ul trasonic flaw detector, employing both a normal probe and a 45 angle

probe, was used to scan the specimen through the region in which the weld was

situated. For each of eight consecutive locations along the surface of the

test section, the ampl itude of the peak response as indicated on the detector

scope for a constant pulse energy was recorded. No attempt was made to

correlate these readings with actual flaw sizes; the method was used simply to

obtain prel iminary data concerning the feasibil ity of using such a device to

detect the initiation of a fatigue crack. The results of the ultrasonic

detection study are presented and briefly discussed in Appendix A.

If the radiograph showed no fatigue crack in the test section the

procedure above was then repeated. This process continued until such time as

a crack became evident, after which radiographs and ultrasonic readings were

r e pe a te d ate ve r y 500 to 1000 c y c 1e sun til the s pe c i me n fa i 1e d .

5.3 Initiation of Fatigue Cracks

This phase of the investigation consisted of a study, using radio-

graphic detection, of the initiation of fatigue gracks in the 3/4 in. thick

HY-80 butt welded specimens. The test specimens contained various types of

weld defects; these defects were generally placed in the second weld pass and

were located approximately midway between the specimen faces. The objectives

of the investigation were twofold: (1) to determine whether, at a given stress

level, some direct correlation exists between the type or size of weld defect

and the initiation of a fatigue crack; and (2) to determine if there is some
 -17-

simple relationship between the cycles to initiation and the specimen "test
;',
1 i fell. It should be clearly understood that, for the purposes of this study,

initiation is taken as the first appearance of a visible crack (macrocrack)

as shown by the radiographs using the procedure outl ined above. The significance

of correlating initiation with test 1 ife rather than cycles to failure will be

discussed in the section on crack propagation.

For the five specimens tested at 50.0 ksi, the 1 ife for fatigue

crack initiation varied from a low of 1,400 cycles for specimen Z-99 to a

maximum of 9,000 cycles for Z-94. A similar wide variation in cycles to init-

iation was exhibited by the specimens tested at 0 to +50.0 ksi. The 1 ife for

initiation ranged from 2,500 cycles for Z-lOO (an admittedly very poor specimen

containing large initial weld defect) to 106,000 cycles for Z-98. Data for

all the specimens appears in Table 5.1.

The large amount of scatter at both stress levels indicates that the

existence of weld defects has a pronounced effect on the resistance of a

specimen to crack initiation. It is assumed that the geometric configurations

of these defects create high stress concentrations that produce plastic

deformations in the vicinity of the flaws, thereby leading to the eventual

initiation of fatigue cracks. That such is the situation was evident from an

examination of the sequence of radiographs taken of each specimen. Visible

fatigue cracks would almost invariably originate at the extremities of a weld

defect, al though not necessarily at the largest defect. This was in marked

contrast to 3/4 in. thick sound weld specimens (with reinforcement removed)

where fatigue cracks frequently had their origins outside the test section.

Test 1 ife - defined herein as the number of cycles necessary for a crack
to propagate to the surface of the test specimen.
 -18-

The type and size of weld flaw leading to fatigue crack initiation

for each specimen is compared with cycles to initiation in Fig. 5.3. For

those specimens tested at 0 to +50.0 ksi, the general behavior appears to be

toward decreased cycles to initiation as size (length) of weld defect

inc re as e 5 . At 50 . 0 ks i, howe ve r, t his te n den c y i s not e v ide n t . I twa s not

possible to determine the geometry at the extremity of the weld defects from

the single radiograph taken at each life.

From these tests, it appears that no single weld defect parameter

in itself controls the fatigue crack initiation. Rather, it is probably the

combined effect of a number of factors that creates the condition necessary

to initiate cracking under a given stress cycle. These factors include:

(1) size and geometry of weld defects; (2) number and type of defects in

weld area; (3) location of defects in specimen (i .e., proximity to other flaws

or specimen surface); (4) orientation of defects with respect to direction of

loading; and (5) the residual stresses in the weld metal. Without more data,

the inter-relationships between the critical variables defies analytical

description, thus pointing to the necessity of continued extensive study of

all factors using a statistical approach. The problem is further compl icated

by the difficulty encountered in trying to create isolated defects of various

types, shapes, and sizes in a specimen.

As might be anticipated from the wide range of cycles to crack

initiation in specimens at both stress levels, there was no consistent relation-

ship between the cycles to initiation and the test life. At 50.0 ksi, the

ratio of cycles to initiation to test 1 ife varied from 0.264 for specimen

Z-99, to 0.833 for Z-94 (Table 5.1). For the specimens tested at 0 to

+50.0 ksi, the variation in this ratio ranged from 0.462 for specimen Z-80

to 0.922 for Z-102. The cracks in the majority of specimens, however,

 -19-

initiated at more than half. the test 1 ife, with the average for all specimens

being 0.69. Thus, on the basis of these tests and the crack detection procedure

used, the number of cycles necessary to propagate the fatigue cracks to the

specimen surface was found to be less than 50 percent of the test 1 ife.,

5.4 Crack Propagation

The process of propagation of a crack initiating at an internal weld

defect in the 3/4 in. thick plate specimens can be divided into two reasonably

distinct regions. The first stage begins with the initiation of the crack

and continues until the crack approaches or actually intersects the specimen

surface (test 1 ife). It is during this period that the specimen experiences

true "fatigue crack" propagation, with the crack growing radially from the

defect and in a plane normal to the direction of loading. Thereafter, the

mode of propagation in the joints changes to one in which the crack progresses

along planes obl ique to the loading direction, exhibiting the so cal led

"shear-type" crack growth behavior illustrated in detail in a earl ier report(4).

The rate of propagation is rapidly accelerated during this latter stage and

ends with complete fracture within a relatively few cycles. This increased

rate of growth is due in large part to the higher stresses created as the

remaining net specimen section required to resist the load is decreased. The

extent of both modes of crack propagation can be easily distinguished by

visual observation of a specimen fracture surface, such as that shown in

Fig. 4.8.

The importance of test 1 ife rather than cycles to complete failure

as a measure of fatigue behavior now becomes apparent. It is the test 1 ife

that most accurately represents the 1 imit of extension of a true fatigue

crack in a plate specimen of the type used in this study, thus provid,ing a
 -20-

consistent bound for evaluating the radiographic data regarding such crack

growth. It was the purpose of this phase of the project to investigate both

the fatigue and shear-type crack growth mechanisms and to cor/elate duration

and rate of propagation with test-stress conditions. A discussion of the

observations is presented below.

The number of cycles of propagation from initiation to test 1 ife,

for the specimens tested at 50.0 ksi, varied from 1000 cycles for specimen

Z-97, to a maximum of 4,500 cycles for Z-96. There was an apparent general

trend but no clearly defined consistent ratio between duration of fatigue crack

growth and specimen 1 ife. Rather, Fig. 5.4 shows that the number of cycles

of propagation falls within a fairly narrow band regardless of the total 1 ife

of the specimen. This observation is even more pronounced for the tests

performed at 0 to +50.0 ksi, when all of the tests are considered. However,

as shown in Fig. 5.5, the cycles of fatigue crack propagation for the current

series of tests ranged from 2,700 for specimen Z-lOO to 14,000 for Z-98, and

show an increase in duration of crack propagation with an increase in test

1 ife, although the increase is not great. The specimens from the previous

tests had greater durations of propagation even though their total 1 ives were

lower.

That the duration of fatigue crack propagation should be approxi-

mately constant at a given stress level may be reasoned by a closer observa-

tion of the crack appearance. Although the geometric configurations of

internal weld defects do vary considerably (thus affecting initiation), once

a fatigue crack forms and begins to extend radially outward from the defect,

the geometry at the tip of the crack becomes similar in most cases. The

specimens then continue to undergo essentially constant fatigue crack growth

up to the point of intersection with the specimen surface, when the mode
 -21-

changes to the shear-type mentioned above. Since the crack-initiating defects

were located near the center of the weld in the specimen and all specimens in

the group tested were 3/4 in. thick, it appears reasonable that they should

exhibit similar propagation times as, indeed, is substantiated by the infor-

mation presented in Figs. 5.4 and 5.5. The sl ight increase in duration of

propagation with test 1 ife may be due to such factors as variation in flaw

location, the residual stresses in the weld, etc.

The second objective of this study on propagation was to investigate

the actual rate of crack propagation, during both the normal fatigue and shear-

type growth stages. The results of measurements from the radiographs are

presented in Figs. 5.6 and 5.7 for tests performed at 50.0 ksi and 0 to

+50.0 ksi respectively. The data are referred to test life, thus different-

iating the two separate stages of propagation. Although the scatter band for

crack length prior to the test 1 ife appears to indicate a uniform rate of

growth, tracing the data for an individual specimen tends to indicate a

stepped pattern of propagation, especially in the tests at 0 to +50.0 ksi.

This system of al ternating periods of growth and dormancy or near dormancy

has been reported in the 1 iterature(6,7) and appears to be fairly common

behavior in the early stages of fatigue crack propagation. The limited number

of observations in these tests, however, precludes any estimate of the number

or duration of these dormant periods.

In the zero-to-tension tests this step-type growth pattern continued

in some specimens for a time beyond the appearance of the crack on the specimen

surface. Thereafter, the growth rate accelerated rapidly and failure usually

occurred within few additional cycles. Figures 5.6 and 5.7 do~ however, show

a considerable range of cycles from test life to failure~ Although no

conclusions have been drawn concerning this variation in duration of crack

 -22-

propagation at each stress cycle, two possibil ities are offered. First, the

planes that the crack fol lows after reaching the specimen surface are often

dictated by the location of other weld flaws in the near vicinity. The crack

will tend to grow toward these defects, especially if they, too, have

initiated fatigue cracks, If no such additional defects exist, the crack

often passes near or into the specimen heat affected zone and fol lows that

1 ine to failure. Depending upon the crack path taken, the crack will progress

through the test section at a slower or more rapid rate. A second possibil ity

would be the location of the propagating crack in the specimen cross-section.

A crack situated toward the edge of the specimen would tend to induce bending

as well as axial stresses in the rema'ining net section, thus increasing the

wedging action at the tip of the crack and further stimulating the rate of

propagation.

5.5 Conclusions

Based on the study reported herein, the following conclusions have

been drawn regarding the initiation and propagation of fatigue cracks in 3/4

in. thick HY-80 transverse butt wel~ed specimens containing weld defects:

1. At a given stress cycle, there is no consistent number of cycles

for the initiation of a macroscopic fatigue crack; nor is there a constant

ratio between cycles to initiation and test 1 ife (appearance of crack on

specimen surface). The number of cycles required to initiate such a crack is

apparently strongly dependent upon several critical parameters including:

a) type, quantity and geometry of internal weld defects~ and b) location of

defect in weld (whether in a tensile or compressive residual stress zone), and

orientation with respect to di rection of loading.

2. After ,the initiation of a fatigue crack at or in the vicinity

of a weld defect, the duration of its propagation to test 1 ife falls within a
 -23-

fairly narrow range of cycles for a given stress level. This period defines

a region of IIfatigue crack" propagation. In the present series of tests this

crack propagation ranged from 1,000 to 4,500 cycles for specimens tested at

50.0 ksi and from 7,000 to 14,000 cycles (excluding Z-lOO) for those tested

at 0 to +50.0 ksi. In most cases the mechanism of early crack propagation

appeared to fol low a stepped pattern of intermittent periods of growth and

dormancy rather than a constant or uniformly varying rate.

3. Beyond the point at which the crack reaches the specimen surface,

the increased stresses near the tip of the crack and throughout the entire

remaining net section cause a rapid increase in the rate of crack growth lead-

ing to failure within a relatively few cycles. During this period of crack

propagation,the path of the crack usually progresses along planes obl ique to

the direction of loading. This pattern then continues until complete fracture

of the specimen takes place.

 -24-

VI. METALLURGICAL STUDIES OF WELDED HY-80 STEEL

6.1 Introductory Remarks

In the previous study(4) on the effect of weld flaws on the fatigue

behavior of butt welded joints in HY-80 steel fatigue crack propagation was

observed. These fatigue cracks propagated either along the fusion 1 ine~ in

the weld metal or in the base metal. Frequently the crack propagation in

the base metal was observed along metal-inclusion boundaries. In a few cases,

the inclusions were melted near the fusion 1 ine leaving small hot cracks in

that region.

Because of the above observations the inclusions were suspected

of causing early fatigue failures when they produced hot cracks in the base

metal near the fusion 1 ine of the weld. These inclusions were tentatively

identified as sulfides because of their low mel ting point and their metal 10-

graphic appearance.

The metallurgical studies of weldments in HY-80 steel in this report

involved three main objectives:

(a) To develop quantitative relationships between the size of

sulfide stringer inclusions and their contribution to hot cracking in a

we 1dment.

(b) To determine the separate contributions of the sulfides and

the stresses during welding on the initiation of cracks in the weld deposit

and heat affected zone of the base metal.

(c) To determine the efficacy of changes in welding procedure on

the red uc t ion 0 r eli min a t ion 0 f hot c r a c kin gin the de po site d we 1d me tal .

After reproducible hot crack production in butt welded HY-80 steel

was attained, fatigue tests were performed on these joints.

 -25-

6.2 Experimental Procedures

Restraint was appl ied to the HY-80 steel during butt welding in

order to develop a number of hot cracks. Welding procedure P80-11018 was

0
used to make the joints. The 30 double "V II grooves for normal butt welded

samples were cut in 3/4 in. and ]-1/2 in. thick HY-80 steel plates, with a

1/8 in. spacing between the plates at the root. Two 4~1/2 in. x 9 in.

plates were then placed in a 17 in. square block of 2 in. thick plain carbon

steel as shown in the diagram in Fig. 6.1. Both HY-80 steel plates were

then fil let welded into the plain carbon steel plate at both surfaces to

produce maximum restraint conditions during the butt welding. The

longitudinal restraint was decreased to some extent by fillet welding the

HY-80 plates only in the direction parallel to the long axis of the butt

weld. After the butt weld was completed, it was allowed to remain under

restraint for a minimum of 24 hours after the butt welding. The fi1 let

welds were initially checked by dye penetrant for possible surface cracking

but none was observed.

Since the hot cracks could not be determined by x-ray radiographY9

the samples were examined by metallurgical sectioning, pol ishing, and

etching. The length and position of the hot cracks were recorded. TL_
I flt!

sample was then repol ished and the hot cracks were again observed in order

to check the first observation.

A diagram of the test plate positions in the original HY-80 steel

plate is shown in Fig. 6.2. The test plates were cut into 9 in. wide strips

for test purposes. The butt weld was approximately 8 inches long in the 9 in.

"VII groove? but only the middle 6 inches of each butt weld was examined for

hot cracks. Metallographic sections were cut from the weld as shown in

Fig. 6.3 after each butt weld was held in restraint for a minimum time of
 -26-

24 hours. The optimum observations of hot cracks were made at a magnification

of 400 diameters,

6.3 Material Description

The HY-80 steel was received in the quenched and tempered condition

and was welded in this condition for the metallurgical studies.

The chemical composition of each of the two HY-80 heats used in

this investigation is 1 isted in Table 2.2 as given by the "Mil 1 Report. 1I An

analysis for manganese, phosphorus and sulfur in each plate near the 3/4 in.

plate surface and at its mid-thickness is given in Table 601. An average of

the resul ts found in Table 6.1 will be used to represent the G-67, G-68, G-69,

and Z-87 plates, The G-69 plate was cut down from a 1-1/2 in. thick plate by

removing 3/8 in. from each side in order to obtain a 3/4 in. thick plate of

a higher sulfur composition. The Z-87 plate had been rol led down to its

3/4 in. thickness at the mill.

The G plates reprepent a higher sulfur content and were taken from

the bottom center of the ingot (plate A-5) as shown in Fig. 6.2. Since the

main rolling direction was known, a reasonably perpendicular cross-section

of the inclusions could be obtained. A direct measurement of the inclusion.'s

width and thickness for the G-69 and Z-87 plates was made at 300x.

6.4 Distribution and Size of Sulfide Inclusions

An average size for the inclusions was recorded by measuring the

widths and thicknesses of a number of inclusions in a section of the plate

transverse to the rolling direction at 300 diameters magnification. The

widths of the inclusions in the Z-87 plate were greater than those in the

G-69 plate on the average~ but the thickness of the inclusions in the Z-87

sample was less than those of the G-69 sample. If an average area of
 -27-

inclusion perpendicular to the roI I ing direction is calculated~ assuming

a rectangular area, the average inclusion area in the Z-87 plate was 165 square

microns compared to 83 square microns for the G-69 plate inclusion area.

Table 6.2 1 ists the resul ts of this study and shows the difference in the

inclusions in the G-69 and Z-87 plates. Another approach taken to describe

the inclusions was to show the size distribution of a number of inclusions

measured within a selected area of the pol ished and etched sample. These

histograms of size distribution are shown in Figs 6.5 and 6.6. The histogram

in Fig. 6.5 shows two or more peaks in the width measurements because some

inclusions were discontinuous and they were measured separately. The only

difference observed between the width in the Z-87 and G-69 histograms is that

the inclusions are pinched off more frequently in the Z-87 samples.

The cracks were usua11y in stringer bands consisting of discontinuous

inclusions. An inclusion was usually located near the fusion 1 ine where the

crack was formed. The cracks that are associated with inclusions do not

have any characteristic size or shape. The histograms of inclusion width

and thickness measurements in samples G-69 and Z-87 also indicate that the

geometric size or shape of the inclusions do not affect the crack frequency

that is given in Table 6.3.

6.5 Discussion

The hot cracks referred to in this report are the small cracks

radiating from the inclusions that have partially melted near the fusion

1 i neon the bas e me tal sid e . In some instances isolated flaws were observed

in the weld metal such as slag inclusions, porosity, cracking along fingers

of the weld metal and a few relatively large cracks not associated with

inclusions. These wil'/ be discussed later under the heading Ilmetallographic

observations of weld flaws. 11

 -2S-

a Hot Crack Geometry

The hot cracks were associated with inclusions because they were

found either radiating from an inclusion near the fusion 1 ine or they were

in 1 ine with stringer laminations on the base metal side of the weld. Examples

of both these conditions are shown in the photomicrographs of Figs. 6.7 and

6.So By sectioning the welds in both the transverse and longitudinal directions

the plane of the hot cracks was observed. The longitudinal section was along

the long axis of the weld and parallel to the fusion 1 ine. The hot cracks were

found to extend less than 0.025 mm from the fusion 1 ine into the base metal

in most cases. Most hot cracks extended from 0.025 mm to D~D5 mm along the

long axis of the weld, The plane of the fractured surface of the hot crack

was pa ra 11 e 1 to the ro 11 i ng direct ion.

b. Effect of Restraint of the Butt Weld on Hot Cracks

Unrestrained butt welds were made by welding procedure PSO-ll01S-D

on 1-1/2 in, thick HY-SO steel samples and no hot cracking was observed in

me tal log rap h i c sec t ion s 0 f the s e we 1d s . I nan a t t em p t to c rea tea dd i t ion a 1

shrinkage stresses in the butt welds 1-1/2 in. x 1 in. x 1 in. blocks were

placed at the ends of the original 9 inch long butt weld double "V" groove in

order to prevent transverse shrinkage. few hot cracks were observed in this

sample. A summary of the above resul ts is 1 isted in Table 6.3.

The butt welding during full restraint described in the experimental

procedure section was successful in causing numerous hot cracks in the G-69

samples. Partial restraint of the butt weld, resulting from only fillet

welding the HY-SO plates in a direction parallel to the long axis of the butt

weld, also caused hot cracks. The amount of cracking per transverse sample

surface examined can be found in Table 6.3.

 -29-

c Effect of Sulfur Concentration on Hot Cracking

In the above tests only the 1 -1/2 in. thick G-69 plates of higher

sulfur content (0.0155% S) were examined. The following tests were conducted

on 3/4 in. thick plates butt welded by the same procedure. One G-69 plate was

cut down from its 1-1/2 in. thickness to a 3/4 in. thick plate and welded to a

rolled 3/4 in. thick Z-S7 plate with a lower sulfur content (0.011% S). The

unrestrained 3/4 in. plate 1 isted in Table 6.3 was a 1-1/2 in. thick HY-SO

steel plate cut down from plate G-67 by machining 3/4 in, from only one side.

No restraint was used in welding the cut down G-67 plate and only a few hot

cracks were observed in the numerous samples prepared (Table 6.3). The

restrained, 3/4 in. thick sample in Table 6.3 was made from a 3/4 in. thick

plate cut down from a 1-1/2 in. G-69 plate by machining off 3/S in. of the rol led

surface from both sides. This plate was welded to a rol led down 3/4 in. thick

Z-S7 plate. The striking result of the butt welded G-69 and Z-S7 plate was

that almost all the hot cracking was on the high sulfur side of the fully

restrained 3/4 in. butt weld.

d. Length of Hot Cracks

Tab 1e 6. 4 1 i s t s the ave rage hot c r a c k 1eng t h sob s e rve din the t ran s -

verse sections for all the welding conditions previously discussed for both the

1-1/2 in. and 3/4 in. thick samples. Most of these small hot cracks are about

0.025 mm long. Large cracks from other types of welding flaws are not included

in the values given in Tables 6.3 and 6.4, The relatively consistent hot crack

length can be related to the I imited depth to which inclusions melt beyond the

fusion 1 ine during the welding.

e. Metal lographic Observations of Weld Metal Flaws

Large cracks, apparently not related to inc1usions~ were observed in

rare instances and an example of this type of cracking is shown in Fig. 6.9.
 -30-

Cracks were also observed to initiate from the periphery of slag inclusions,

porosity (Fig. 6.10) and Ilfingers" of weld metal in the base metal (Fig, 6.11).

These cracks were usually larger than the hot cracks and Table 6.5 1 ists the

samples in which these defects were found along with their crack lengths,

f. Inclusion Study

A more exact description of the inclusions was desired because of

the association of hot cracks with inclusions in the HY-80 steel microstructure.

Sulfides were assumed present from the results of a sulfur print which was

rnadeo Methods of extraction for the inclusions resulted in extracted particles

with an index of refraction between 1.35 and 1.63c This result was based on

Becke 1 ines observed around the edges of a large number of extracted stringers

immersed in 1 iquids of known indices of refraction. Sulfides must have an index

of refraction of two or better. it was concluded that sulfide inclusions were

not successfully extracted. X-ray diffraction patterns of the extracted

particles matched the ASTM pattern for iron sil icate. The sulfides were probably

dissolved in the solvent used for extracting inclusions since they were not

observed on the remaining sample or filtered from the solution.

g. Effect of IIButtering l1 on the Frequency of Hot Cracks

Two sections of the 1-1/2 in. thick G-69 plate were joined under ful 1

restraint by butt welding using welding procedure P80-11018-~ However, before

one section was placed in the full restraint apparatus it was Ilbuttered 'i

(coated with Ell018 weJd metal by weaving the electrode over one side of the

The excess Ilbuttered il weld metal was machined off leaving a layer
0
30 scarf.)

of weld metal approximately 1/8 in. thick. The 1-1/2 in. thick sections were

then butt welded, sectioned and observed, The unbuttered side of the weld

had an average crack frequency and length similar to other 1-1/2 in. samples

welded in full restraint. The only cracks observed on the 'Ibuttered 'l side of
 -31-

"the butt weld were near the root of the weld where i'buttered" weld metal had

not been placed. This clearly demonstrated that hot cracks can be decreased

to a low level by "buttering" the HY-80 steel before using regular welding

procedures provided the remaining buttered layer of weld metal is thick enough

to prevent the melting of inclusions in the base metal.

A microhardness survey (Fig. 6.12) in a transverse section of the

"buttered" and normal G-69 plate shows the different Diamond Pyramid Hardness

(DPH) in the Heat Affected Zone (HAZ) of the two plates and al so in the

remaining "buttered " weld metal left after machining. The maximum hardness in

the HAZ of the "buttered" section was lower than maximum hardness in the

unbuttered HAZ of the HY-80 steel butt weld. A hardness survey of the HAZ of

the buttered scarf is shown in Fig. 6.13. This hardness survey shows the

effect of the large heat input caused by the extensive weaving motion of the

electrode during the buttering operation.

h. Effect of Hot Cracks on Fatigue Life

A G-69 plate, 1-1/2 in. thick, was machined down to four 3/4 in. thick

plates by removing 3/8 in. from both sides giving a thinner plate of higher

sulfur content (0.0155% S), These plates were butt welded in full restraint to

make two samples. The samples were then tested in a method similar to a

regular 3/4 in. thick fatigue test samplen These samples were tested in 0 to

+50.0 ksi tensile stress fatigue cycles. One sample reached a fatigue 1 ife of

170~000 cycles and the other 280,000 before an attachment weld at the end of

each sample broke. The fu11y restrained welds containing hot cracks were not

broken. The attachment weld was repaired for the sample that reached 280,000

cycles and the test was continued. The attachment weld broke twice before a

total of 430,000 cycles was reached. The G-69 sample welded in restraint had

not broken, even after 430,000 cycles of stress of 0 to +50.0 ksi.

 -32-

The order of magnitude calculation given below illustrates the

probable number of hot cracks contained in the 4 in. wide 3/4 in. thick fatigue

sample used in the test just described. From measurements reported in Table 6.3

each transverse section of a cut down 3/4 in. G-69 sample welded in restraint

should contain one (1) hot crack on the average. If it is assumed that each

hot crack extended along the long axis of the weld for 20 mils. (0.020 in.)

and no hot cracks overlapped along the long axis of the weld, then the number

of hot cracks is equal to 4 in./0.020 in. = 200 cracks in a 4 in. wide, 3/4 in.

thick test sample. The 430~000 cycle fatigue 1 ife of the fully restrained weld

sample approaches the fatigue 1 ife of the unwe1ded plate. Thus, the hot cracks

along the fusion 1 ine do not appear to have an effect on zero=to=tension fatigue

1 ives at this stress level.

Another 1-1/2 in. HY-80 steel sample was welded in restraint caused

by two 15 in. long plain carbon steel bars 3 in. square that were fillet welded

along the sides of the G-68 plate. The 1-1/2 in. thick plate was fillet welded

as if it were a 3/4 in. thick plate. The restraining bars were removed after

twenty-four hours and the entire plate was cut down to a 3/4 in. fatigue

sample. This sample was fatigue tested at a stress level of 50.0 ksi. It

failed after 13,000 cycles compared to a good weld fatigue 1 ife of about

60,000 cycles at the same stress level. The sample was sectioned and pol ished

across the fatigue fracture region. Fig. 6.14 shows the fractured region after

13,000 cycles. The arrows indicate the cut of the transverse section shown in

Fig. 6.15. The transverse section is near the center of the fatigue fracture.

The entire crack is in the weld metal in this region. Hot cracking in the HAZ

could not be responsible for the fatigue crack center. An investigation of

this center of the fatigue crack and centers of fatigue cracks in the weld metal

indicates that these centers are probably weld porosity.

 -33-

In a region of the same transverse section shown in Fig. 6.15 only

on the opposite side of the weld about 1/4 in. from the fracture, porosity was

observed. This pore was located near the root pass of the weld and very close

to the fusion 1 ine. This is a common location for small pores in the 3/4 in.

and 1-1/2 in. thick HY-80 steel butt welded plates. This pore was too small

to be observed in an x-ray radiograph of the plate even with reinforcement

removed. Two sma11 cracks were observed radiating out from each end of the pore

as shown in the metallographic section in Fig. 6.16. The cracks were nearly
0 0
parallel, within 4 , with one crack direction making an angle of 85 to the
0
fatigue stress di rection and with the other crack making an angle of 87.5 to

the stress direction. Further proof that these were fatigue cracks rather than

shrinkage cracks was the fact that at a high magnification the cracks were

fine, close fitting rather than open3 smooth shrinkage separations.

6.6 Conclusions

The following conclusions are based on resul ts obtained from two

commercial heats of HY-80 steel;

1. Restraint during the welding of HY-80 steel promotes hot cracks

in the HAZ of HY-80 weldments. The hot cracks are on the average 0.025 mm in

length for the two heats investigated.

2. Higher concentrations of sulfur in HY-80 appear to promote hot

cracking. Hot cracks were observed for 0<0155% S, but practically no hot

cracks were observed for 0.011% S HY-80 steel weldments. However, the size

(width and thickness) of the inclusions does not appear to be related to the

hot crack frequency from the measurements of inclusion size made in this

investigation.
 -34-

3. IIButtering" prevents hot cracking. "Buttering 'l al so causes a

sl ightly lower maximum hardness in the HAZ of the "buttered" HY-80 steel

compared to unbuttered HAZ.

4. Hundreds of hot cracks (micro-cracks) in the sample do not appear

to reduce the fatigue 1 ife to any great extent for weldments tested in tension

from to+50.0 ksi for the 0.0155% S HY-80 steel welded in full restraint. The

long axis of the weld was perpendicular to the axis of tension and the hot

cracks propagated in the general direction of tension causing a small stress

raiser factor.

5. In one case porosity was observed to be a nucleus for fatigue

cracking in the weld metal of the HY-80 steel sample G-68 subjected to 13,000

cycles of a reversal stress level of -50:0 ksi to +50.0 ksi. Porosity in the

weld metal was bel ieved to be responsible for the early fatigue failure of

the G-68 sample welded in restraint.

 -35-

VII. SUMMARY

The following is a brief summary of the result~ of the various studies

reported here in. Ali mj ted number of tes ts were cond'Jcted and many of the

conclusions are based on a relatively small amount of data .. N\~vefthe~ess, there

are a number of observations that are considered to be very signific~nt and are

briefly summarized herein.

7.1 Fatigue Behavior of HY-I00 Steel

Fatigue tests carried out on plain plate and butt welded specimens

of HY-l00 indicate:

1. The axial fatigue strength of ~he HY~100 steel plate was not

greatly different from that of the'HY-80 plate previously tested under a

zero-to-tension stress cyc1e~,'

2. Under a zero-to-tension loading, transverse butt welded joints in

HY-IOO were found to have a lower fatigue resistance than comparable joints

in HY-80. Furthermore, the HY-100 joints appeared to be more sensitive to

the presence of internal flaws.

7.2 Weld Defect Study of HY-80 Steel

From the fatigue tests carried out on butt welds with intentional

intern?} defects it may be concluded that:

1. For the range of intentional defects studied, the effect on the

fatigue strength is independent of the type of defect and the amount of

defective area.

2. At 100,000 cycles the fatigue strength of the members with intentional

flaws is about the same as that of the as-welded joints with the reinforcement

in place (no intentional defects).

 -36-

3. At shorter 1 ives (less than 100,000 cycles), it appears that the

intentional internal weld defects in specimens with weld reinforcement removed

are more important than the external geometry of some as-welded joints, while

the external weld geometry appears to be more important at the longer 1 ives.

7.3 Crack Growth Study

Radiographic techniques were employed to determine the number of

cycles to crack initiation and subsequent propagation of fati~ue cracks in

butt welded joints of HY-80 steel. Because of the lack of sensitivity and

difficulty of interpretation of the radiographs, there is some uncerteinty as

to the time of initiation. Nevertheiess, c3 nUn)ber of conclu,>ions can be drawn

on the basis of the test results:

I. For a given stress cycl~ there appeared to be no consistent number

of cycles for the initiation of a macroscopic fatigue crack; rather, the

initiation appears to be a function of the type and geometry of the internal

weld defects, the location of the defect with respect to the cross section of

the weld, and the residual stresses.

2. Upon initiation of a fatigue crack, the duration of propagation was

found to fall within a relatively narr~~ range of cycles for a given level of

appl ied stress.

3. On the basis of the crack detection procedure used in these tests,

the number of cycles necessary to propagate the fatigue crack to the specimen

surface was found to be less than 50 percent of the total 1 ife.

7.4 Metallurgical Studies

A number of metallurgical studies have been conducted to evaluate

the effects of inclusions and welding restraint upon the qual ity of weld metal
 -37-

in butt welds of HY-80 steel e The results of these studies may be summarized

as follows:

1. The restraint during \'lfelding of HY ... 80 steel produced hot cracks in

the heat affected zone of the weldment.

The degree of hot cracking appeared to increase with an increase

in the sulphur concentration of the steel.

2. "Buttering" of the beveled plates helped to prevent hot cracking in

the heat affected zone.

3. The hot cracks (microcracks) in the \'Ieldments did not appear. to

reduce the fatigue 1 ife of the members to any great extent.

7.5 General Summary and Conclusions

On the basis of the results of the various tests reported herein

several general observations may be made concerning the behavior of HY-80 and

HY-I00 steels and welded joints in these steels. These conclusions are as

fol lows:

1. Based on the average values of fatigue strength at 100,000 cycles

shown in the fol10\'Jing tabulation plain plate members of !-IY-80 and HY ... I00

are found to have fatigue strengths considerably above that of mild structural

steel. However, the increase in fatigue strength for the butt welded joint

in HY-I00 is much smaller than that for the HY-80 steel.

Steel F100,OOO' Ksi (Zero-to-tension)

Plairi Plate Transverse Butt Weld*

Mild structural steel (~~TA-A7) 48.0 32.3

HY-80 65.6 57. )

HY-100 68.6 41.4

"k As-wel ded, re i nforcement on

 -38-

The as-welded joint efficiencies under axial loading are seen to

range between 60 and 87 percent for the HY-I00 and HY-80 materials respectively.

2. Radiographic evaluations in HY-80 demonstrate that the test joints

~"'ere wi thi n Navy spec i f i cat ions 1 imi ts for poros i ty and slag. HO\-1ever II when

the \oJe 1d re i nforcement was removed from members hav i ng fl aV'/s with j n the

specification 1 imits, their fatigue strength at 100,000 cycles was no better

than that of sound as-welded butt joints ",lith the reinforcement in place .

.3. Prel iminary results of ultrasonic measurements for crack prop3gation

and detect ion in HY-80 suggest that th i s detect i on method if it can bl~ full y

developed, may provide earl ier crack detection than radiographic ex~mination.

HO\'Jever, this can only be real ized through extensive developments in the

operational techniques.

4. The microcracks observed in the butt YJeld'~d Joints of HY ... SO had

less effect on the fatigue resistance of t:le members than trie porosity and

slag inc1usions permitted by present inspect;on specifications.

 -39-

BIBLIOGRAPHY

1. Sahga1, R. K. and Munse, W. H., "Fatigue Behavior of Axially Loaded

Weldments in HY-80 Steel ," Univers ity of 111 inois, Department
of Civil Engineering, Structural Research Series 204,
Septembe r 1960.

2. Hart ma nn, A. J. and Mu n s e, 'vi. H., I I Fat i g ue Be h a v i 0 r 0 f We 1de d J 0 i n t s

and We1dments in HY-80 Steel Subjected to Axial Loadings,"
University of Illinois, Department of Civil Engineering,
Structural Research Series 250, July 1962.

3. Z i mme r ma n, J ~ E., Jr., and Mun s e, W. H., I IFat i g ue Be h a v i 0 r 0 f

Defective Welded Joints in HY-80 Steel Subjected to Axial
Loading," University of III inois, Department of Civil Engineering,
Structural Research Series 252, July 1962.

4. Hartmann, A. J., Bruckner, W. H., Mooney, J. and Munse, W. H.,

"Effect of Weld Flaws on the Fatigue Behavior of Butt-Welded
Joints in HY-80 Steel," University of Illinois, Department of
Civil Engineering, Structural Research Series 275,
December 1963. .

5. Bureau of Ships, U. S. Navy, "Fabrication, Welding and Inspection

of HY-80 Submarine Hu1Is," NAVSHIPS 250-637-3, January 1962.

6. L i u, H. W., IIC rack P ropagat ion in Th in Me ta 1 Shee t Unde r Repea ted

Loading," University of !11 inois, Department of Theoretical and
App] ied Mechanics, Report No. 155, December 1959.

7. Lip sit t, H. A., For be s, F. W. and Ba i rd, R. B., II Crack Pro p a gat i on i n
Cold Rolled Aluminum Sheet," Proceedings, American Society for
Testing and Materials, 1959.
 TABLE 2.1

PHYS I CAL PROPERT lIS OF BASE METAL

(Data Suppl ied by Manufacturer)

Propert ies in the Longitudinal Direction

Heat De sign at ion Thickness Yi e 1d ";', Tensile Elong. Reduction Charpy V-Notch
Numbe r (inches) Strength Strength in 2 inches in Area ft~lbs.
(ks i) (ksi) (pe rcent) (percent) @ -120 o F

69S344 G 1-1/2 88.5 108.3 22.6 69.2 92

I
..J::::.
N31543 Z 3/4 90.0 109.2 25.3 70.4 136 0
I

N15423 HY 3/4 110.0 127.5 23.0 71 .1 83

I',
0.2 percent offset
 -41-

TABLE 2.2

CHEMICAL COMPOSIT!ON OF BASE METAL

(From Mil J Report)

Chemical Content Heat Numbe r

(pe rcent) 69S344 N3l543 N15423

C 0.16 0.17 0.20

Mn 0,33 0.28 0.30
P 0.021 0.010 0.010
S 0.019 0.009 0.014
Si 0.26 0.27 0.21
Ni 2.86 2.29 3.00
Cr 1.61 1.31 1.67
. Mo 0.48 0.32 0.50
Cu 0.06 0.11
 TABLE 3.1

RESULTS OF FAT I GUE TESTS OF HY-l 00' PLA I N PLATE SPEC I MENS
(Zero-to-Tension)

...}(
Computed Fatigue Strength~ ksi
Spec i men Stress Life Locat ion
Numbe r Cycle (cycles) of ,1\,', F50 ,000 F100 ,000 F200 ,000 F500 ,000
(ksi) Fracture

HY-4 o to +80.0 46~400 a 78.3 63.5 51.7

HY-6 o to +80.0 57,100 f 83.1 67.6 54.8

I
HY-7 o to +76,5 81 ,200 f 88.4 71.7 58.4 .p,.
tv
I

HY-3 o to +50.0 1180> 100 f 73.5 59.7 48.4 36.8

HY-2 o to +50.0 ~~53,100 a 81.1 66.1 53.6 40.7

HY-5 o to +50.0 !546,300 a 102.5 83.2 67.5 51.3

Average 84.5 68.6 55.7 42.9

if,
k=0.30l
,.,,',
a: failure initiated at r~dius of test section
f: failure initiated at mill scale surface near radius of test section
 TABLE 3.2

RESULTS OF FATIGUE TESTS OF HY-IOO TRANSVERSE BUTT WELDS IN

THE AS-WELDED CONDITION
(Zero-to-Tension)

II:;
Computed Fatigue Strength, ksi
Spec i men Stress Life Loca t ion
Numbe r Cycle (cycles) of F20 ,000 F50 ,000 F100,000 F200 ,000
(ksi) F rac tu re-/c;lnlc;

HY-: 9 o to +80.0 45 700 b ( 4 5 , 1),1,,1(,

HY- 8 o to +80.0 5~400 b ( 4 7 . 7 )'1,,,,,

I
~
HY~ 10 o to +80.0 8,200 b (56.3 )'I,;\- W
I

HY-13 o to +50.0 24,100 b (53 .8 )''n\'

HY-ll o to +50,0 61,500 c 77.7 54.3 41.3 31.4

HY-12 o to +50.0 61 ~ 800 c 77.9 54.4 41.4 31.5

HY-14 o to +30.0 2123500 c -------- 53.1 40.4 30.8

HY-15 o to +30.0 240,600 c -------- 55.2 42.4 32.2

Ave rage 77.8 54.2 41.4 31.5

,'\
k = 0.394
,~.};.!I,

not included in average; fatigue strengths were calculated only for sound welds
i',/'~i'\

b: failure initiated in weld metal

c: failure initiated at edge of weld reinforcement
 TABLE 3.3

SUMMARY OF FATIGUE TESTS OF HY-80 AND HY-IOO MATERIAL

(Zero-to-Tension)

a) 3/4 In. Thick HY-80 Material (1)

Computed Fatigue Strength, ksi

S pe c i me n Type Surface Condition k
F
F20 ,000 F509000 F100sOOO F200 ,000 2,000,000

Plain Plate As-Roll ed 0.201 91.0 75.6 65.6 57. I 35.8

Transverse Butt Weld As-Welded 0.279 69.3 57.1 45.0 23.8
Transverse Butt Weld Reinforcement Removed 0,229 80.7 69.0 58.9 34.8 I
~
~
I

b) 3/4 In. Thick HY-IOO Material (2)

Plain Plate As-Ro 11 ed 0.301 84.5 68.6 55.7

Transverse Butt Weld As-Welded 0.394 77.8 54.2 41.4 31.5

c) 1-1/2 In. Thick HY-80 Material (3)

Plain Plate As-Rol led 0.149 89.9 78.5 67.9 60.9 43.6
Transverse Butt Weld As-'We I ded 0.340 77.3 56.3 44.6 35.0
Transverse Butt Weld Reinforcement Removed 0.146 82.0 71.2 64.3 58.2

( I) results reported in Ref. 4

(2, ) results of this study
(3 ) re~ults reported in Ref. 2
 TABLE 4.1

RESULTS OF FATIGUE TESTS OF TRANSVERSE BUTT WELDS iN

1-1/2 IN. HY-80 STEEL
(Intentional Porosity in Weld)

_'~

Computed Fatigue Strength,A ksi

Spec i men S t res s Life Location
Numbe r Cycle (cycles) of ...1......1...
I .... , .. F20 ,OOO F50,OOO F100 ?000
(ksi) Fracture

G-82 o to +70.0 6,300 b 55.2 45.7 39.8

G-89 o to +70.0 6,700 b 55.9 46.4 40.2
G-84 o to +70.0 7 ~ I 00 b 56.6 46.9 40.6
G-85 o to +70.0 7)900 b 57.8 47.9 41.5 I
.p.
U1
G-83 o to +70.0 10,900 b 61.8 51.2 44.4 I

G-81 o to +50.0 22,800 b 51.3 42.7 36.9

G-87 o to +50.0 39~nOO b 57.5 47.6 41.4
G-86 o to +50.0 53.,600 b 61.0 50.7 44,0
G-88 o to +50.0 56:1 000 b 61.8 51.2 44.4

Average 57.7 47.8 41.5

,',
k = 0.210
,'"I,
b: failure initiated in weld metal
 TABLE .4.2

RESULTS OF DEFECT EXAMINATION OF TRANSVERSE BUTT WELDS IN

1-1/2 IN. HY-80 STEEL
(Intentional Porosity in Weld)

-;I~/I\
Defective Area, Percent
Spec i men Stress d Radiographic Description
e ,I.

Numbe r Cycle Based on Based on Rating" of

( i 11.
(ksi) Radiograph Fracture Surface Defect

G-82 o to +70.0 0.035 0.027 0.080 P one c I us te r

G-89 o to +70.0 0.051 0.248 0.115 P one cluster
G-84 o to +70.0 0.053 0.045 0.285 P one cluster I
-J::>.
G-85 o to +70,0 0.065 0.050 0.250 P one cl uster (J)
I

G-83 o to +70.0 0.039 0.219 0.058 P one c 1us te r

G-81 o to +50.0 0.046 0.088 0.129 P two clusters; offset in
the direction of stress
G-87 o to +50.0 0.067 0.219 0.334 F one c 1 us te r
G-86 o to +50.0 0.064 0.180 0.184 P one cluster
G-88 o to +50.0 0.046 0.158 0.282 F two clusters; offset in
the di rection of stress

...),
based on radiographic requirements of Navy Specifications(5)
...,',...,1,
effective diameter of largest pore in porosity clusters
 TABLE 4.3

RESULTS OF FATIGUE TESTS OF TRANSVERSE BUTT WELDS IN

1~1/2 IN. HY-80 STEEL

(Intentional Slag Inclusions in Weld)

-'.
Computed Fatigue Strength," ksi
Spec i men Stress Life Location
Numbe r Cycles (cycles) of , , .......... 1,.,
... ... F20 ,000 F50~000 FIOO,OOO
(ksi) Fracture

G-102 o to +70.0 6,700 b 55.6 4~). 9 39.7

G-95 o to +70.0 8 800
j b 58.9 48.5 42.0
G-96 o to +70.0 10,900 b 61.7 50.8 44.0 I
,.p.
G-98 o to +70.0 11,300 b 62. I 51.2 44.2 '-I
I

G-93 o to +50.0 26,400 b 52.9 43.7 37.8

G-103 o to +50.0 28,700 b 53.9 44.5 38.5
G-92 o to +50.0 42~800 b 58.7 4B.4 41.8
G-91 o to +50.0 58 500
j b 62.7 51 .8 44,7
G-90 o to +50.0 98,800 b 69.9 57'.8 49.B
Ave rage 59.6 49.2 42.5

...,',
k = 0.210
it,,':.:
b: failure initiated in weld metal
 TABLE 4,4

RESULTS OF DEFECT EXAMINATION OF TRANSVERSE BUTT WELDS IN

1-1/2 IN. HY-80 STEEL
(Intentional Slag Inclusions in Weld)

Defective Area, Percent

Specimen Stress Radiographic Description
Numbe r Cycle Based on Based on Rat i ng,l, of
(ksi) Radiographs Fracture Surface Defect

G-l02 o to +70.0 l.49 1.00 F 2 defects

G-95 o to +70.0 0.022 0.025 P
G-96 o to +70.0 0.340 0.394 F 2 defects
I
G-98 o to +70.0 0.349 0.040 P +:0-
ro
I
G-93 o to +50.0 0.163 P defect not visible
on X-ray
G-I03 o to +50.0 0.688 0.027 F
G-92 o to +50.0 0.271 0.054 P
G-91 o to +50.0 0.258 0.076 F 2 defects
G-90 o to +50.0 0.041 0.055 P porosity + slag

../~
based on radiographic requirements of Navy Specifications(5)
 TA BLE 4.5

RESULTS OF FATIGUE TESTS AND DEFECT EXAMINATION OF TRANSVERSE BUTT WELDS IN

3/4 IN. HY-80 STEEL
(Intentional Porosity and Slag)

Defective Area, Percent

Specimen Stress Life Compu ted Radiographic Type of
Numbe r Cycl e (cycles) F ,S\i'~
Based on Based on Rating De fec t
50 sOOO
Radiograph Fracture Surface
(ksi)

Z-94 50.0 14,800 37.3 0.72 0.61 F slag

Z-96 50.0 145300 37.0 0.40 0.12 F slag
';'~
Z-56 50.0 14,200 37.0 0.025 0.11 P poros i ty
I
Z-74-,',: 50.0 13,400 36.5 0.14 0.19 P slag .p..
ill
I
Z-95 50.0 8,900 33.0 0.047 0 P slag
Z-99 50.0 5,800 29.9 0.75 0.39 F porosity + slag
Z-97 50.0 5,300 29.1 0.37 0.20 F porosity
'k (l )
Z-77 47,4 4,600 26.8 0.64 0.71 F slag
~/;

Z-57 30.0 145,900 38.8 0.046 0.18 P porosity

,'r
Z-78 30.0 90 ~ 100 34.5 crack 0.14 F slag
~/::
Z-75 30.0 78,800 33.5 crack 0.13 F slag

,'\
data from Ref. 4
i'\'i~\
assuming K = 0.24
(I) subjected to bending
 TABLE 5.1

RESULTS OF FATIGUE CRACK INITIATION AND PROPAGATI'ON STUDY OF

TRANSVERSE, BUTT WELDS IN 3/4 IN. HY-80 STEEL
CONTAINING INTENTIONAL WELD DEFECTS

{I ) {2) {3) {4) {5) {6) {7~ {8 )

Spec i men S t res s Cycles to Test Life to Rat i 0 Initial Defect Length Type of De fec t
Numbe r Cycle Start of Life(b) "Fa i I ure" (c) Col.3/ Col ,4 Measured on Fracture
(ksi) Crack Surface (inches)
(a) (cycles)
Growth
(cycles)

Z-94 50.0 9,000 10~800 14,800 .833 .18 slag

Z-95 50.0 4,800 6,900 8;;900 .696 .05 slag
Z-96 50.0 8,600 13,100 14,300 .656 .04 slag
Z-97 50,0 2,800 3,800 5~300 .737 .27 porosity
Z-99 50.0 1 ,400 5,300 5~800 .264 .05 poros i ty I
(J1
0
Z-98 o to +50.0 106,000 120;;000 130,500 .883 .07 porosity I

Z-IOO o to +50.0 2,500 5,200 17 ~ 300 ,481 2 - .05 "ea. drilled holes
(:45)(d) (lack of fusion)
Z-101 o to +50.0 90,000 99,400 104,400 .905 2 - .05 ea. d rill ed ho 1es
Z-102 o to +50.0 82,100 89,000 111,500 .922 II - .05 ea. dril led holes
Z-70(e) o to ~50. 0 43,700 52,300 55,600 .836 .14 slag
Z-79~e) o to +50.0 17~000 26?300 32,900 .646 .16 slag
Z-80(e) o to +50.0 11,400 24,700 27,200 .462 .31 slag

(a)
estimated as 1/2 no. cycles between last x-ray showing no fatigue crack and first x-ray showing crack
(b)
estimated as 1/2 no. cycles between last observation showing no crack on specimen surface and first
observation showing surface crack
(c)
defined in Chapter I (usually less than complete fracture)
(d)
fatigue crack initiated ~t lack of fusion, not at"drilled holes
(e)
data from earlier report(4)
 -51-

TA BLE 6.1

MANGANESE, SULFUR, AND PHOSPHORUS CONTENT OF

THE HY-80 STEEL PLATES
( in Pe rcen t)

Sample Samp 1e Manganese Phosphorus Sulfur

;'~
Position

G-69 0.28 0.014 0.016

0.29 0.012 0.015

2 0.30 00012 0.016

2 0.28 0.010 0.016

Ave rage 0.288 0.012 0.0155

Z-87 0.26 0.008 0.011

0.28 0.008 0.010

2 0.30 0.008 0.013

2 0.28 0.008 0.010

Average 0028 0.008 0.011

,':
Position ( 1) was near the surface and position (2) was at
the mid-thickness of the plate.
 -52-

TABLE 6.2

AVERAGE SIZE OF SULF!DE INCLUSIONS PER

NUMBER MEASURED
(microns)

S amp 1e Plate Average Average

Thickness !nc1usion
~'-
I nc1 us ion",
Width" Thickness

G-69 1-1/2 in. 32/21 2.59/21

Z-S7 3/4 in. 110/25 1 .5 /25

sec t ion cut pe r pe nd i c u1art 0 the r 0 11 i n g

direction and across the thickness of the
plate

TABLE 6.3

NUMBER OF HOT CRACKS PER TRANSVERSE SECTION

OF HY-SO STEEL EXAMINED

Thickness of Steel Plate

Sample Test Type of
Numbe r Plate Restraint 1-1/2 in. 1-1/4 in. 1 in. 3/4 in.

G-67 weak 2/7 0/12 0/5 0/5

transverse
2 G-67 none 0/6 0/2 2/9
3 G-69 both transverse 21/6
and longitudinal
4 G-69 longitudinal 14/3
restraint relaxed
5 G-69 both t ransve rse 6/7
Z-S7 and longitudinal
 -53-

TABLE 6.4

AVERAGE LENGTHS Of HOT CRACKS

(M. ill i me te rs)

../~
Thickness of HY-80 Steel Plate
Samp 1e
1-1/2 in, 3/4 in.

0.02
2 0.02
3 0,03
4 0.02 0.04

same samples as shown in Table 6.3

TABLE 6.5

LENGTHS OF HOT CRACKS NOT ASSOC!ATED WITH INCLUSIONS

(M ill i meters)

."
Thickness of HY-80 Steel Plate
S amp] e
1-1/2 in, 3/4 in.

3A 13
crack a 0.08
crack b 0.11
3Al 0.20
4D5 0.26
2D10 0.50

All samples were from the 1-1/2 in. thick

G-69 test plate, Thel-l/2 in. thick
samples were cut down on one side in
2D10 and on both sides 4D5.
 4'- a"

5"
1
I

----r-
-------- 4"
3" __-----:---.i.

I"
I T6 Dio. Holes
t

(a) Plain Plate

r' d~1\7"Rod.
I i \
ED CD e e ! I Gl e (9 ~
G (1) -+(01+---

e e e e

L -_ _ _ _ _ _ _ _ _ _ _ _ ~_ _ _ _ _ _
D See flgs.2.2 8 2.3 for weld detail,;
~~------~------------~I-~
t1"
4

(b) Transverse Butt Weld

~ 5" ~
11
I \\ 7 Rod.
I -----e---e~-~---e~.

. CD e $----.--- - - - ------ -- ---f'-1- ------..-~-_e--.e_ - -

(9 (D' e e

See fig. 2.4 for weld details.

J *

(c) Transverse Butt Weld

\

FIG. 2.1 DETAILS OF TEST SPECIMENS.

 ,.. 4" Test Section ... ,

J I ... 5
I I
:3 - I

3'
I
I
!
T 2 I
I
!I
I I ... 4
. 6 I
I
:,
I. 6" Continllous Weld Passes .. I
Root opening 13 in.
Arrows indicate direction of weld ing.
Surface of plate adjacent to weld cleaned by grinding before welding.

Pass Electrode size, in. Current, amps. Rote of travel, in. / min.

I i 130 5
32 ~

2 c 140
~ '- 5

3 2- 230 8
16

4 l 220 7
16

5 l 210 7
16

6 l 210 1
16

Voltage :21 Volts

Polarity: D.C. Reversed
0
Preheat: 150 F
Electrode: MIL 11018
0
I nterpass Temperature: 200 F (Maximum)
Heat Input: 40,000 Joules/ in. (Maximum)
All welding in flat position
Underside of pass I glound before placing pass 2

FIG.2.2 WELDING PROCEDURE P80-IIOI8 .... A

(Transverse Butt Welds in HY- SO)
 l ! ....
5
I I
3 I
I
I I
I .. I
I
2 I
I
I I ... 4
I
I I
6 I

~l- I 6" ContInuous

I~~.--------------------------~
Weld Passes i

Root opening t In
Arrows Indicate direction of welding

Surface of plate adjacent to weld cleaned by grinding before welding

Pass Electrode size, In Current, amps Rate of travel, In. / mrn

I i 130 5
32
..5
2 32 140 5
-
3 l 230 8
16

4 .l 220 7
16
_.- .-
5 3 210 7
IS
--~-~
6 ~ 210 7 I
16
I
VOl tage . 21 Volts
Polarity. o C. Reversed
0
Preheat 150 F
Electrode MIL 11018
0
I nterpass Temperature 200 F (Maximum)
Heat Input 40,000 Joules/ In (Maximum)
All welding in flat position
Underside of pass I ground befGfe pioc,ng pass 2

FIG.2.3 WE LDING PROCEDURE PIOO-1I0IS-J

(Transverse Butt Welds in HY-IOO)
 60
, .. 2't" Test Section _I

,I I
17,19,21,23
11,13,15
J I
I
7,9
1 ~

5
I

2 I
I
J
J I
I I
3
4 I I
I I
6 I
8,10 I I

I I
12,14,16
18,20,22,24 I
I
I
I

4" Continuous Weld Posses'

Roo t opening ~ in. Arrows indicate direction of welding.

Surface of plate adjacent to weld cleaned by grinding before welding.

Pass Electrode size, in. Current", amps. Rate of travel, in./min.

5
1 32
140 4
3
2 16 230 6

5
3 32 160 4

3
4-16 16 230 6

3
17-24 16 200 5
Vol toge: 21 Volts
Polarity: D. C. Reversed
o
Preheat: 200 F
Electrode: MIL 11018
interposs Temperature: 200F (Maximum)

Heat Input: 50,000 Joules I in. (Maximum)

All welding in flat position.
Underside of pass I bock-gouged with oi r arc before pass 3.

FIG.2.4 WELDING PROCEDURE P80-IIOI8-D

(Transverse Butt Welds in HY .... 80)
 Lever

Dynamometer

Specimen

FIG.2.5 ILLINOIS FATIGUE TESTING M-ACHINE AS USED

FOR AXIAL LOADING OF WELDED JOfNTS
 0) BASE METAL. (230X MAGNIFICATION)

b) HEAT AFFECTED ZONE. (230X MAGNIFICATION)

F IG.3.1 TYPICAL PHOTOMICROGRAPHS OF TRANSVERSE

BUTT WELD IN 3/4 IN. HY... 80 MAT ERIAL.
 0) BASE METAL. (230X MAGNIFICATION)

b)H'EAT AFFECTED ZONE. (230X MAGNIFICATION)

FIGs3.2 TYPICAL PHOTOMICROGRAPHS OF TRANSVERSE

BUTT WELD IN 3/4 IN. HY-IOO MATERIAL.
 100 --,....---
I
-- 1 - - - - -1 - - ' - - - I
~.... 1'-~ I --
__ --.. ---- j

--
80 ~-

N~
-~ ~.... .. ,

.
1;::--

60 ~ .-
~
~- -- - --- --
~
'-'

.. -.J. V
/
r- r-k=O.20

.. r0- t--
~~
--
1- .... ~
40
(J) k = 0.30-./
.........
r--- -- I"'"- __ ..
'-
.lC:

(J)
(J)

-
Q)
....
( J)

E 20
::::J
E
)(
0
~

10
_. Stress Cycle: Zero-to- Tension
I

8 - 0
3"
HY-IOO
4
- 3"
----- "4 HY- 80
6 r-
,

See Fig. 2.10 for specimen detoi Is

5 -' I I 1-i--sts- 1 ,
20 40 60 8 IDo 200 400 60 000

Cycles To Failure, In Thousands

FIG. 3.3 RESULTS OF FATIGUE TESTS OF AS - ROLLED HY-IOO PLAIN PLATE SPECIMENS.
 NOTE: FRACTURE DID NOT START AT CORNER.

FIG.3.4 FRACTURE SURFACE OF AS-ROLLED HY ... IOO

PLAIN PLATE SPECIMEN.
 100 -
-lw
j
--- ...... ..--- ---,
-- ~ ~
I

80 r - - - - - f--.

-
-.-~---

"
I

--1--
....... I

V-- k = 0.297
60 --r-- -_.
~
"' ........
......
.... r--. - I

W ~ 1-,0,.
~
r---- ......... ........ /
................
JI
~ ~ .........
~ r-....
I

.........
(f)
40
k = 0.394 r-- V "" ~ ""'" .........
r--....... ....
~
..::c. ~ .......

~
(f)
(f)

-G,)
b.

( J)

E 20
:J
E
)(
0
~

W: Indicates failure in weld

10
- Stress Cycle: Zera - to - Tension
8 - 3"
0 "4 HY-IOO
-
6 r- ---- ~"
4
HY-80
See Fig. 2.lb for specimen details
5. _1 I 1 I I I I
2 4 6 8 10 20 40 60 80 100 200 600

Cycles
,
To Failure, In Thousands

FIG. 3.5 RESULTS OF FATIGUE TESTS OF HY-IOO TRANSVERSE BUTT WELDS IN THE
AS-WELDED CONDITION.
a) FAILURE INITIATED AT EDGE OF WELD REINFOR CEMENT.

b) FAILURE INITIATED IN WELD.

FIG.3.6 TYPICAL FRACTURES OF TRANSVERSE BUTT

WELDS IN 3/4 I.N. HY-IOO MATERIAL.
 100 - ....
~----~-r~~~TT~--I-

1=U~~~~~IT~
I _

r-

80 -

o~

-
.c
t J')
c:

-
Q)
'-
V>
GJ
60

I
\- .L~~
'0
'r\'{-~...--
2. __i
,..-'--"'-
_
-

::J ~

-
tJ')

o
LL
-~~
....-
~"'F"
--,""
I
-
c: 40 1___+--t-.::--.11--:-:-\-\-::;'(-=_
I L
\\(O~ ~ ...... --~
1"'-

-..... _r _
c: ~ \n __ ....... 1 ~.... ~

-
o
( .)
- - I-- .-, , ___ /
.....- -,- _>--r-

~ 1~~~_~:~~=4~-+-tt1----t-~----r-~--
::J '-
-0

~ \n. Ll'i~O
Q)
Q:
~~. / ,!.-.,....... _ _
~
20
L_ _
... f'" --- I I

~/ './ -f.---~ - - ---I--'"" I-"" I

1
Stress Cycle:
Z
1ero
to Tension
1
o
20 40 60 80 100
... ,
200 400 600 800 1000 2000 4000 6000

Cycles To Failure, In Thousands

'FIG. :3.7 REDUCTION IN FATIGUE STRENGTH FROM AS - ROLLED PLAIN PLATES DUE TO A
TRANSVERSE BUTT WELD IN THE AS-WELDED CONDITION.
 100r.------~----~--,_--r_Ir~_,~------_r----,_------._--_,----r__.--.__._._.------II----u_------r----.---,---,

80 1 I ~ I 1- i':: I I I I I I ~~

60
4--I-k = 0.21

40
(I)
..lI::

~1-4-
(I)
(I)

-....
Cl)

en
E 20
:::J
E
)(
0
~

101 I I
t Stress Cycle
8[ Zero - to - Tension

6+-
See Fig. 2.lc for specimen details
5'
2
I 4I 6
I I I I I~
8
I
20 4
~
6
~ I
80
I I
100
I
200
I
400
g
600

. Cycles To Failure, In Thousands

FIG. 4.1 RESULTS OF FATIGUE TESTS OF 1-1/2 IN. TRANSVERSE BUTT WELDS CONTAINING
VARIOUS AMOUNTS OF INTENTIONJ~L POROSITY (REINFORCEMENT REMOVED).
 a) POROSITY IN WELD.

b) SLAG IN WELD.

FIG.4.2 TYPICAL FRACTURES FOR 1-1/2 IN.

TRANSVERSE BUTT WELDED JOINTS
CONTAINING INTENTIONA INTERNAL
WELD DEFECTS.
 100

-c:
Q)

E
Q)
0
to-
O 90
-...
'0::wc:
-
..c.
'i
en 80
3.2 I
Q)

~
,

"'0
c
~
0
(f)
70 (Butt Welds - As Welded)
-... 0 0
0 0
0
0
0 .. u u
8 )
o 0
lJ...
60
\6-

-0

c
Q)
0
"- E
"'0
Q)
>
0
r-
Stress Cycle
Zero - to - Tension
Note: Diameter is based on maximum
0

OJ OJ
a.. a:: pore on fracture surface.
--~~ I I I I
50
o 0.01 0.02 0.03 0.04 0.05 ~ 0.06 0.07 0.08

Effective Defect Diameter, in.

FIG. 4.3 PERCENT FATIGUE STRENGTH AT 100,000 CYCLES VS. EFFECTIVE DEFECT
DIAMETEH OF POROSITY IN 1-1/2 IN. MATERIAL.
 100

CI) 80
~
~

CI)
CD

~
U (Butt Welds - Reinforcement Removed)
o
o 60
o.
o
o
....o 00
(Butt Welds - As . Welded)
0
:ED 40 1)~ <) 0 .....,
C
CD
C
....
t...
(f)

QI
:::s

-
~
0-.

20 ------ ~- 1 - - -r--'

Stress Cycle
Zero - to - Tension
Note: Area is based on the defect
area shown on radiograph.
o
.02 .04 .06 . .08 10 20 .40 .60 .80 1.0 2.0 4.0 6.0

Percentage of Defective Area, % (A red )

FIG. 4.4 COMPUTED FATIGUE STRENGTH AT 100,000 CYCLES vs. DEFECTIVE AREA DUE TO
INTENTIONAL. PCROSITY IN 1-1/2. IN. MATERIAL.
 100r-=r--------r----~----~--~~--,_~~--------~-----~------~r_----r_--_r--_r--,_~--r_~-------~----~:-------~----_r----r__,

80

40+--
en
.lII::

U)
U)

-I"+-
-~
00
E 20t--
I I I I
::J
E
)(
o
~

,0~----~--1---L-JL-L-1-L4------~--_+----_4--~~_t--t_i_i_t_r_----t_~1l----_t--~--1I~
t- Stres s Cycle:
8-1- Zero - to - Tension
I-

6+-
See Fig. 2.lc for specimen details
.i_~_ I __ I J _ I I
52 . 4" 6' S" . 10 20 40 60 80 100 200 400 600

Cycles To Failure, In Thousands

FIG. 4.5 RESULTS OF FATIGUE TESTS OF 1-1/2 IN. TRANSVERSE BUTT WELDS CONTAINING
VARIOUS AMOUNTS OF INTENTIONAL SLAG (REINFORCEMENT REMOVED).
 /00 -

CI) 80
.lI:

CI)
.
CD

~
U (Butt Welds - Reinforcement Removed)
oo 60
o flO

o
o lA
IU
....o b.
(Butt Welds - As' Welded)
fA
.c:. b. b.
~ 40 L
l.
c
Q) 0 ~
....'-
en
OJ
:::J

-
.~
~ 20

Stress Cycle
Zero - to - Tension
Note: Area is based on the defect I
_~grea show~~__ !adiog~aph.
o - - '----~~~ -~ '---------------~ '-------- I

.02 .04 .06 .08 ./0 .20 .40 .60 .80 1.0 2.0 4.0 6.0

Percentage of Defective Area, % (A red )

FIG. 4.6 COMPUTED FATIGUE STRENGTH AT 100,000 CYCLES vs. DEFECTIVE AREA DUE TO
INTENTIONAL SLAG INCLUSIONS IN 1-1/2 IN. MATERIAL.
 100 .,

(Butt Welds - Reinforcement Removed)

en 80 (Butt Welds - As' Welded)
~
...
en
Q)

~ "-
U
bt. J.P.
o
o 60 Do
~

Do
o f.
bt.
__
ci
N 0 ~

-C

-
.t::.
en
c
40

-Q)
'-
UJ
W
::J

-
.2'
0
LL
20

Stress Cycle
Zero - to - Tension
Note: Area is based on the defect
area show~ on ~CJdi9grc:Je~~___ ~
o -~-

.02 .04 .06 .08 .10 ..20 .40 .60 .80 1.0 2.0 4.0 6.0

PercentClge of Defective Area, % (A red ,>

FIG. 4.7 COMPUTED FATIGUE STRENGTH AT '20,000 CYCLES vs. DEFECTIVE AREA DUE TO
INTENTIONAL SLAG INCLUSIONS IN 1-1/2 IN. MATERIAL.
 ELEVEN DRILLED HOLES IN A ROW.

FIG.4.8 FRACTURE SURFACE OF 3/4 IN. TRANSVERSE

BUTT WELDED JOINT CONTAINING ARTIFICAL
INTERNAL WELD DEFECTS.
 100 ~

en 80
~

en
.. ,-

CD

~
U

oo 60
o
ci (Butt Weld - As . Welded)
lO
....o
.c.
1; 40 U
IE:: 0 /::,. L ~
Q)
/::,.
....
'-
(J)
0
GJ
::J
A

-
.~
~ 20
o Porosity /::,. Slag
Stress Cycle
Complete Reversal
Note: Area is based on the defect
area shown on radiograph.
o
.02 .04 .06 .08 .10 .20 .40 .60 .80 1.0 2.0 4.0 6.0

Percentage of Defective Area, % CAred)

FIG. 4.9 COMPUTED FATIGUE STRENGTH AT 50,000 CYCLES VS. DEFECTIVE AREA DUE TO
INTENTIONAL WELD DEFECTS IN 3/4 IN. MATERIAL.
 100

en 80
~

en
..
Q)

~
U (Butt Weld - Reinforcement Removed)
o
o 60
o
o
o ~

...
o
~

00 0 ~ fA 1 - r--'"
(Butt Weld - As' Welded)
.s:::. ~ ( o _~
en 40
,.....

~
fj L:.
c C
~
...
en
0

cu
:s

-
.~
If 20
o Porosity ~ Slag
Stress Cycle
Zero - to - Tension
Note: Area is based on the defect
area shown on radiograph.
o
.02 .04 .06 .08 .10 .20 .40 .60 .80 1.0 2.0 4.0 6.0

Percentage of Defective Area, % (A red )

FIG. 4.10 COMPUTED FATIGUE STRENGTH AT 100,000 CYCLES \IS. DEFECTIVE AREA DUE TO
INTENTIONAL WELD DEFECTS IN 1-1/2 IN. MATERIAL.
 a) EQUIPMENT AND LEAD SHIELD IN POSITION.

b) SPECIMEN AND X-RAY FILM HOLDER.

FiG. 51 SETUP FOR RADIOGRAPHIC STUDY OF

CRACK PROPAGATION.
 0) EQUIPMENT SHOWING NORMAL AND 45 ANGLE PROBES.

b) EQUIPMENT IN POSITION FOR RECORDING.

FIG. 5.2 SETUP FOR UL TRASONIC STUDY OF

CRACK PROPAGATION.
 (I)
c:
" Stress Cycle (/)
(1)
Type of c
Q)
E 50.0 ksi 0-50.0ksi Defect
100 E
0

en
(1)
a.
.\
0
~
Slag
0
Q)
Q.

c: "
I
Porosity
Drilled Holes
(J)

~
0
0 I
\\ Lack of Fusion .JIC:
0
0
'- 80 '-
U
(1)
\ u
W
::I ::I
.Q'
+- \ (0 to + 50.0 ksi) .~
0 +-
0
LL
~

0 60
\ LL
~
0

0
c:
\ c
.2

.
+- +-
.2 .Q

\
+-
:!:
c (I) c
~

\
.EO
"0
40
" 0
10 -
0

(/)
".
(/)
~LO

" "
0 "O~
en + 0
.
::::lI
0
L:.
0
+-
'\, I
I
(/)

::1
oLO
0
..... 0 I L:.
..... -H
c
- 20 u 5
c
-
"
0
0

~J
en 'U
~/). .............. 'U
(1)

-
QlI (/)
- +- (1)
W
U (/)
to-

----
(/)
>. w (,)
U ..... /). >.. w
u .....
- -I-- .
o
o .10 .20 .30 .40 .50

Initiol Length of Weld Defect, inches

FIG. 5.3 CORRELATION BETWEEN INITIAL LENGTH OF INTENTIONAL

WELD DEFECT AND INITIATION OF FATIGUE CRACKS.
 8 --
~'pecirnen Symbol
en Z-94- 6.
~ Z-95 0
E ~ Z -96 V
ou
.... Z -97 0
LL ...... Z-2.2, 0
c 0 6
o en
:;: "0
o c:
0\ 0
o en
a. :J
00
.... ..c: \I
a..t-
oX c: 4
u '-
o
U
.... Q)
!t:
~ ..J 0
~-
.- en
~
- .....Q) 2
...... 0
o -
c c
o 0
:;::;:
o
.....0
:J
-
.'!:
o c
o
o 2 4 6 8 10 12 14 16

Specimen Test Life., in Thousands of Cycles

FIG. 5.4 COMPARISON OF CYCLES OF FATIGUE CRACK PROPAGATION TO TES LIFE IN 3/4 IN.
HY-80 TRANSVERSE BUTT WELDED SPECIMENS TESTED AT + 50.0 KSI.
 40
Specimen Symbol
(f)
Q)
Z - 98 /::,. I
I
Z-IOO
E ~ z -101 "
f--------+---t----------I ---------t----
ou 0
'-
lL. ..... Z-102 0 I

o 30 f--------.J
c
o
:;:
CJ)
"0 Z-70 * A
o c
0\ 0
o CJ) Z-79 "* I I I
a. :::J
o 0
'- ..c.
a.. I-
l-80*
-* Data

from prevIous study
..lit: C 20
g I
u'- Q)
~

--
....J

Q)
:J i~
0\
:;:
o
-CJ)
Q)

. -- --L --- .-- ----

--- ---- --
lL. I- 10
II ........
.....0
-
i-

--- ------
0
0
c

--
0
0
'-
c
0
0
V-- --- --- ----
:::J
0 C
o
o 20 40 60 80 100 120 140 160

Specimen Test Life, in Thousands of Cycles

FIG. 5.5 COMPARISON OF CYCLES OF FATIGUE CRACK PROPAGATION TO TEST LIFE IN 3/4 IN.
HY-80 TRANSVERSE BUTT WELDED SPECIMENS TESTED AT 0 TO 50.0 KSI. +
 __-------------------~----~---------~--------,_--------_r~-~~~m~~~~t~~~~~~~
(J)
w
.c.
u
c

~
"..
(/)
.c.
a.
I I
o I ~
r...
0' en
~2 Q.)

"0 I-
o I 201--------+---------+-
0::
c Region f Fatigue .. ,.:Y<~~~~
o Crack P opagation. .r,l:~P;-~;Pf

- ~:~:,~~,;P)ffip,1~~ftt
"0
W
o
.~ .801 , ..
"0
c:
..:.:: F_ ;>

u Cra
0
r...
U
\4-
0

-
.c:
C"
c:
QJ
.40

-I

o tn' it ',:":'':;,",71- .. \.;z{~ ". ~;'I",;.-..... f ! . ~-~"'( ... -n'",- t'-x"'" I

4000 3000 2000 1000

--+-- 1000 2000 3000 4000

Cycles Prior to Appearance of Surface Crack Cycles Beyond Appearance of Surface Crock

FIG. 5.6 RATE OF CRACK PROPAGATION IN 3/4 IN. HY-80 TRANSVERSE BUTT WELDED SPECIMENS
TESTED AT 50.0 KSI.
 2.00 .
Spec I me n S ym b 0 I
T .. :'t.>n~ .. ':'\.i:,';<~:."h -; .. ! ......, ,I(L ', . . ~~;......" '~ ,:.~ .., .:.... ,~,
, ..
~~''''~)~;.~ ~P.~.::at'~:~:\(~~~.J:'~(~~~:'rt~~f~W<F.lij~W.~~;(l.;;:".;;~h.:9::1.~~it.:}>:':::.
:"";.,.",:),",', ""';":.~. ~'<. t', 0'

~ l;:i~ ~ J_-~_-_--L---
V)
..
..r::.
1.60

1 .;,IIIIJliltllltir'lili!
Z-70~..
A I
~-1
W

C-
o z-79'*. I!
"-
0"
Z- 8 0 * . I (/)
o
.--1-- . ___
---'- W

'0
o
0:;:
120 *" Dota from prevIous study --------111
..- ~,.... ~ r." ' ' ' _~' ~'.'<.~' ~':
,.! ......... .... r''''''''..... _...! .: . k.'''''!t ... <:.i>J\ ... .,,~. ~."'Jf'.\<.'
... ,.0( ..... _ ... . .... 1'.". 1:1I'
i
c F atiguc
o

...o
"0
Q)

,~
"0 .so ....- - - - i -
C

..:
U
o....
U
~
o

-
..r::.
C\
C
w t-F~0~w\n:f
. II S oa~ - Type II t
...J
o Crack l-'r8Do. otlon .

o:t:~~tlt{ }t~~~~&~tiiif~1?;\~%~'/" ~._______~II-________ ..&..._ _ _ _ _ _ _ _..A-_ _ _ _ _ _ _ _.....J._ _ _ _ _ _ _ _- ' ' ' ' ' '_ _ _...II

16000 12000 8000 4000 4000 8000 12000 16000

Cycles Prior to Appearance of Surface Crock Cycles Beyond Appearance of Surface Crock

FIG. 5.7 RATE OF CRACK PROPAGATION IN 3/4 IN. HY-80 TRANSVERSE BUTT WELDED SPECIMENS
TESTED AT 0 TO 50.0 KSI. +
 Section
A AA

Note A
7-2I"

4..L"
2

.',
~
60
~
c::
~ ..
24"
f."

::i:t;
I" ,7 ... )','",".
4- .... -..

L
,'I"t',

~J_'I~:.
l :"':.::~

Note A: Fillet Weld All

Around Except
Across V - Groove,

R. D.: Main Rolling

-12" ~
Direction. ";
CSJ
... : '.
":".~::!,:;.
~

HY - 80 Steel Plate

~ Mild Steel Plate

FIG. 6.1 WELDING JIG USED TO RESTRAIN HY-80 STEEL DURING BUTT WELDING.
 INGOT TOP

t
36" A A-3 A-6

130"
t36" A-I A- 4 A-7

+-L
36" A-2 A-5 Main Rolling Direction
.....

l I~ 144" .1I ~
- - - - - 440"
1 96" .1 j
I
INGOT BOTTOM

FIG. 6.2 MAIN PLATE OF HY-80 STEEL FROM WHICH TEST PLATES G-67, G~68, AND
G-69 WERE OBTAINED.
 LO'IGITUD I NAL AX IS OF WELD :>

TOP VIEW

o *"DISCARJ
IN.
o SAMPLE
~IN.

FIG 6.3 SECTIONING PLAN FOR METALLOGRAPHIC

SAMPLES OF BUTT WELDED HY-80 STEEL.

FIG 6.4 TRANSVERSE SECTION OF A MULTIPASS

BUTT -WELD OF HY-80 STEEL. 2X
 8, i~-----""-------""

71~------------+-------------4--------------r----------~-+-------------1--------------r-------------4-------------4-------------~

61 Z 87 Samples
I
>.
uc: _
~ 41 _
0"
4 W. U. I Micron
~ 3
ll..

fOO 200 300 400 500 600 700 800 900

Width Units - (W.U.)

8.r------------~-------------..,._-----------------~----------------~-------------~----------------~----------------r-------------~~------------_....
71 -1-------------+-------------1-------------~------------_;

61 _
G - 69 Samples
>.
u 511-1-----
C
Q)
:::s 41----..,.-
0"
~ 31 _
ll..
2

0
0 fOO 200 300 400 500 600 700 800 900

Width Units - (W.U.)

FIG. 6.5 HISTOGRAM OF THE WIDTH OF INCLUSIONS IN THE Z - 87 AND G- 69 SAMPLES

OF HY -80 STEEL.
 Z-87 Samples
8~-------+-----

:>.
u 6i-----
c:
Q)
::J 4 T. U. = I Micron
0-
~ 4~--
lJ..

o
o 5 /0 15 20 25

Thickness Units - (T. U.)

14

12

G - 69 Samples

10

:>..
u 8
c:
Q.)
::J
0-
W 6
'-
lJ..

0
0 5 10 15 20 25

Thickness Units - (T. U.)

FIG. 6.6 HISTOGRAM OF THE THICKNESS OF INCLUSIONS IN

THE Z-87 AND G-69 SAMPLES OF HY-80 STEEL.
FIGS.7 PHOTOMICROGRAPH OF A LARG:e: CRACK IN
.THE HAZ AND WELD METAL OF BUTT ... WELDED
HY ... 80 STEEL. 465X

'FIG6.8 PHOTOMICROGRAPH OF THE SAME CRACK

SHOWN ABOVE AFTER 0.009 IN. WAS POLISHED
FROM THE TRANSVERSE SECTION. 200 X
FIG 6.9 PHOTOMICROGRAPH OF A LARGE CRACK

NOT APPARENTLY ASSOCIATED WITH

INCLUSIONS. 400 X.

FIG 6.10 PHOTOMICROGRAPH OF A CRACK ASSOOATED

WITH POROSITY NEAR UNDERCUT REGION

OF TH WE LO. 400 X e
FIG G.U PHOTOMICROGRAPH OF A CRACK ALONG

A ~FINGER~ OF WELD METAL THAT EXTENDS

INTO THE BASE METAL. 150 X.
 600
I
I
I I I 1
I
-,--
13m m
./ /'

500 _L_ 5
...... , ........................... I -
A
I
I 6
I

CI)
CI)
Q)
400
IJ\,,: I
:1
,/
-

l
C
"0
I 0 40

Y
"-

n
0
I
:'Q

a..
E
0
"-
~
300
I,

:Nr--v ----:---- --I

W ~
~

I I

U
"0
c I I
E
0 ....",.,.,..
~ Normal
I I "Buttered" ~~~
0 I I
Base Metal HY-80 HAl Weld Metal HAl Base Metal HY - 80
0 200 F.I, F.I L.
I I
I I

100 -

a
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38

Distance Across Specimen - M. M.

FIG. 6.12 MICRO HARDNESS SURVEY OF 1- 1/2 IN. THICK HY - 80 STEEL BUTT WELD.
 500 T
I I I I I

I
I
I

!(!'J\
..........
I
I
~
40C F.IL. -

~
I
(f) I

~
(f)
Q)
I

~)
C
"0 II Buttered II

~
o
1~F\ HAZ
I
"0
30C
VV VI
A
I
E I
o
~
~
I
I
k --........... ~ V
.........
~ .........
-
0.. I
"0 Weld Metal I Base Metal HY -80 Steel
C 200
o I
E F.IL.
o
o I
I

10C

01
o 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32

Distance Across Specimen - M. M.

FIG. 6.13 MICRO HARDNESS SURVEY OF THE WELD METAL AND HAZ OF "BUTTERED"
HY-80 STEEL SCARF.
FIG 6.14 FRACTURED SECTION OF HY'" 80 STEEL

SAMPLE G-68 WELDED IN RESTRAINT.

fiG 6.15 TRANSVERSE SECTION OF FRACTURED

SAMPLE I NDICA TED BY THE ARROW

IN FIG 6.14.
FIG 6.16 PHOTOMICROGRAPH SHOWING FATIGUE CRACKS

RADIATING FROM POROSllY IN HY-80

STEEL SAMPLE G ... 68 AFTER 13,000 CYCLES

OF i- 50.0 KSI. 120X.
 A-l

APPEND IX A
Ultrasonic Crack Detection Stu~y

The resul ts of the prel iminary ul trasonic crack detection study

are presented in Table A. 1, and plotted in Figs. A.l and A.2 for tests

performed at 50.0 ksi and 0 to + 50.0 ksi respectively. As mentioned in the

report, ultrasonic readings were taken fol lowing each X-ray at various intervals

during the 1 ife of each specimen. The ampl itude of the response as shown on

the oscil 1 iscope screen was recorded for each of eight consecutive locations on

the specimen surface. Only those response readings corresponding to the

location nearest the point of actual crack initiation (as shown by the radio-

graphs) have been recorded in Table A. 1. Generally, the other reading

locations away from the point of initiation showed considerably less fluctuation

unti 1 the crack had progressed for some time. It should also be noted that

the ampl itudes of the responses recorded correspond only to a convenient

setting of the ultrasonic equipment and are not meant to be interpreted as a

measure of true crack dimensions.

Superimposed on each plot in Figs. A. 1 and A.2 are the points where

the radiographs first showed the existence of a fatigue crack. The results

indicate that the ultrasonic system of detection does show promise as a future

means of determining crack initiation in weld specimens. Further experimentation

with the techniques of this detection procedure, however, is necessary to

improve the rel iabil ity of the results.

 TABLE A. 1
RESULTS OF ULTRASONIC STUDY OF CRACK PROPAGATION

Specimen Z-94 Specimen Z-96 Specimen Z-98 Specimen Z-lOl S pe c i me n Z- 102

Cycles Ultrasonic Cycles Ultrasonic Cycles Ultrasonic Cycles UI trasonic Cycles Ultrasonic
read i ng reading read i ng read i ng reading

32 16 26 0 34

2,000 29

1,000 20
0
2,000-
30
35
0
2,300- 28 2,000- 31
97,000 avg. 69,000 avg. 67,200 avg.
4,000 25 2,000 26
103,000 33 76,000 31 77, 100 32
6,000 39 3,600 24
109,000 36 85,000 35 87,000 29 I
8,000 37 6,100 32
N
119 ~ 000 50 95,100 50 91,000 58
10,000 40 7,600 36
121 ,000 50 99,500 54 93,100 58
11,600 56 10,000 36
102,000 57 95,000 50
12,200 53 11,600 41
104,400 62 97,000 56
12,800 53 13,100 41
99,000 49
13,800 48 13,200 46
 Specimen Z-96 45

40

35

Ol 30
c:
"'0
o 25
(l)
a::
(l)
a. 20
o
u
(f)
60~------~------~------~------+-------+-------+-----~15

-
~
o
u

-
(l)

( lJ
o
55

50
Specimen Z - 94

u
c:
o
(J)
45
o
-~

40

35
First indication of
fatigue crack as
shown on radiograph.
25

20~------~------~------~------~----~~------~----~
o 2 4 6 8 10 12 14

Cycles, In Thousands

FIG. A.I RESULTS OF ULTRASONIC CRACK DETECTION

STUDY FOR SPECIMENS TESTED AT 50.0 KSI.
 60
Specimen~ Z - 98
50

0\ 40
c: Avg. Magnitu~
-e--_ .......
"0
o 301 160
Q)

~
0::: Specimen Z -101
Q)
a. 50
o
u
en 40

--
'-
o
u
Q)
Q)
:30

o 651 120
.S? Specimen Z - 102
c
o(J)
55

-o
'-

:J 45
\. First indication of
fatigue crack as
shown on
radiograph.
35

75 85 95 105 115 125

Cycle s, In Thousands

FIG. A.2 RESULTS OF ULTRASONIC CRA,CK DETECTION STUDY FOR

SPECIMENS TESTED AT 0 TO + 50.0 KSI.